A Critical Analysis of Patient Safety Practices - Medical and Public ...
A Critical Analysis of Patient Safety Practices - Medical and Public ...
A Critical Analysis of Patient Safety Practices - Medical and Public ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Evidence Report/Technology Assessment<br />
Number 43<br />
Making Health Care Safer: A <strong>Critical</strong><br />
<strong>Analysis</strong> <strong>of</strong> <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong><br />
Prepared for:<br />
Agency for Healthcare Research <strong>and</strong> Quality<br />
U.S. Department <strong>of</strong> Health <strong>and</strong> Human Services<br />
2101 East Jefferson Street<br />
Rockville, MD 20852<br />
www.ahrq.gov<br />
Contract No. 290-97-0013<br />
Prepared by:<br />
University <strong>of</strong> California at San Francisco (UCSF)–Stanford University<br />
Evidence-based Practice Center<br />
Editorial Board<br />
Kaveh G. Shojania, M.D. (UCSF)<br />
Bradford W. Duncan, M.D. (Stanford)<br />
Kathryn M. McDonald, M.M. (Stanford)<br />
Robert M. Wachter, M.D. (UCSF)<br />
Managing Editor<br />
Amy J. Markowitz, JD<br />
Robert M. Wachter, MD<br />
Project Director<br />
Kathryn M. McDonald, MM<br />
UCSF–Stanford EPC Coordinator<br />
AHRQ <strong>Public</strong>ation 01-E058<br />
July 20, 2001 (Revised printing, indexed)
Preface<br />
The Agency for Healthcare Research <strong>and</strong> Quality (AHRQ), formerly the Agency for<br />
Health Care Policy <strong>and</strong> Research (AHCPR), through its Evidence-based Practice Centers<br />
(EPCs), sponsors the development <strong>of</strong> evidence reports <strong>and</strong> technology assessments to assist<br />
public- <strong>and</strong> private-sector organizations in their efforts to improve the quality <strong>of</strong> health care<br />
in the United States. The reports <strong>and</strong> assessments provide organizations with comprehensive,<br />
science-based information on common, costly medical conditions <strong>and</strong> new health care<br />
technologies. The EPCs systematically review the relevant scientific literature on topics<br />
assigned to them by AHRQ <strong>and</strong> conduct additional analyses when appropriate prior to<br />
developing their reports <strong>and</strong> assessments.<br />
To bring the broadest range <strong>of</strong> experts into the development <strong>of</strong> evidence reports <strong>and</strong><br />
health technology assessments, AHRQ encourages the EPCs to form partnerships <strong>and</strong> enter<br />
into collaborations with other medical <strong>and</strong> research organizations. The EPCs work with these<br />
partner organizations to ensure that the evidence reports <strong>and</strong> technology assessments they<br />
produce will become building blocks for health care quality improvement projects throughout<br />
the Nation. The reports undergo peer review prior to their release.<br />
AHRQ expects that the EPC evidence reports <strong>and</strong> technology assessments will inform<br />
individual health plans, providers, <strong>and</strong> purchasers as well as the health care system as a whole<br />
by providing important information to help improve health care quality.<br />
We welcome written comments on this evidence report. They may be sent to: Director,<br />
Center for Practice <strong>and</strong> Technology Assessment, Agency for Healthcare Research <strong>and</strong> Quality,<br />
6010 Executive Blvd., Suite 300, Rockville, MD 20852.<br />
John M. Eisenberg, M.D. Douglas B. Kamerow, M.D.<br />
Director Director, Center for Practice <strong>and</strong><br />
Agency for Healthcare Technology Assessment<br />
Research <strong>and</strong> Quality Agency for Healthcare<br />
Research <strong>and</strong> Quality<br />
The authors <strong>of</strong> this report are responsible for its content. Statements in the report should not be<br />
construed as endorsement by the Agency for Healthcare Research <strong>and</strong> Quality or the U.S.<br />
Department <strong>of</strong> Health <strong>and</strong> Human Services <strong>of</strong> a particular drug, device, test, treatment, or other<br />
clinical service.
Structured Abstract<br />
Objectives: <strong>Patient</strong> safety has received increased attention in recent years, but mostly with a<br />
focus on the epidemiology <strong>of</strong> errors <strong>and</strong> adverse events, rather than on practices that reduce such<br />
events. This project aimed to collect <strong>and</strong> critically review the existing evidence on practices<br />
relevant to improving patient safety.<br />
Search Strategy <strong>and</strong> Selection Criteria: <strong>Patient</strong> safety practices were defined as those that<br />
reduce the risk <strong>of</strong> adverse events related to exposure to medical care across a range <strong>of</strong> diagnoses<br />
or conditions. Potential patient safety practices were identified based on preliminary surveys <strong>of</strong><br />
the literature <strong>and</strong> expert consultation. This process resulted in the identification <strong>of</strong> 79 practices<br />
for review. The practices focused primarily on hospitalized patients, but some involved nursing<br />
home or ambulatory patients. Protocols specified the inclusion criteria for studies <strong>and</strong> the<br />
structure for evaluation <strong>of</strong> the evidence regarding each practice. Pertinent studies were identified<br />
using various bibliographic databases (e.g., MEDLINE, PsycINFO, ABI/INFORM, INSPEC),<br />
targeted searches <strong>of</strong> the Internet, <strong>and</strong> communication with relevant experts.<br />
Data Collection <strong>and</strong> <strong>Analysis</strong>: Included literature consisted <strong>of</strong> controlled observational<br />
studies, clinical trials <strong>and</strong> systematic reviews found in the peer-reviewed medical literature,<br />
relevant non-health care literature <strong>and</strong> “gray literature.” For most practices, the project team<br />
required that the primary outcome consist <strong>of</strong> a clinical endpoint (i.e., some measure <strong>of</strong> morbidity<br />
or mortality) or a surrogate outcome with a clear connection to patient morbidity or mortality.<br />
This criterion was relaxed for some practices drawn from the non-health care literature. The<br />
evidence supporting each practice was summarized using a prospectively determined format.<br />
The project team then used a predefined consensus technique to rank the practices according to<br />
the strength <strong>of</strong> evidence presented in practice summaries. A separate ranking was developed for<br />
research priorities.<br />
Main Results: <strong>Practices</strong> with the strongest supporting evidence are generally clinical<br />
interventions that decrease the risks associated with hospitalization, critical care, or surgery.<br />
Many patient safety practices drawn primarily from nonmedical fields (e.g., use <strong>of</strong> simulators,<br />
bar coding, computerized physician order entry, crew resource management) deserve additional<br />
research to elucidate their value in the health care environment. The following 11 practices were<br />
rated most highly in terms <strong>of</strong> strength <strong>of</strong> the evidence supporting more widespread<br />
implementation.<br />
• Appropriate use <strong>of</strong> prophylaxis to prevent venous thromboembolism in patients at risk;<br />
• Use <strong>of</strong> perioperative beta-blockers in appropriate patients to prevent perioperative<br />
morbidity <strong>and</strong> mortality;<br />
• Use <strong>of</strong> maximum sterile barriers while placing central intravenous catheters to prevent<br />
infections;<br />
v
• Appropriate use <strong>of</strong> antibiotic prophylaxis in surgical patients to prevent postoperative<br />
infections;<br />
• Asking that patients recall <strong>and</strong> restate what they have been told during the informed<br />
consent process;<br />
• Continuous aspiration <strong>of</strong> subglottic secretions (CASS) to prevent ventilator-associated<br />
pneumonia;<br />
• Use <strong>of</strong> pressure relieving bedding materials to prevent pressure ulcers;<br />
• Use <strong>of</strong> real-time ultrasound guidance during central line insertion to prevent<br />
complications;<br />
• <strong>Patient</strong> self-management for warfarin (Coumadin) to achieve appropriate outpatient<br />
anticoagulation <strong>and</strong> prevent complications;<br />
• Appropriate provision <strong>of</strong> nutrition, with a particular emphasis on early enteral nutrition in<br />
critically ill <strong>and</strong> surgical patients; <strong>and</strong><br />
• Use <strong>of</strong> antibiotic-impregnated central venous catheters to prevent catheter-related<br />
infections.<br />
Conclusions: An evidence-based approach can help identify practices that are likely to<br />
improve patient safety. Such practices target a diverse array <strong>of</strong> safety problems. Further research<br />
is needed to fill the substantial gaps in the evidentiary base, particularly with regard to the<br />
generalizability <strong>of</strong> patient safety practices heret<strong>of</strong>ore tested only in limited settings <strong>and</strong> to<br />
promising practices drawn from industries outside <strong>of</strong> health care.<br />
This document is in the public domain <strong>and</strong> may be used <strong>and</strong> reprinted without permission<br />
except those copyrighted materials noted for which further reproduction is prohibited without the<br />
specific permission <strong>of</strong> copyright holders.<br />
Suggested Citation:<br />
Shojania KG, Duncan BW, McDonald KM, et al., eds. Making Health Care Safer: A<br />
<strong>Critical</strong> <strong>Analysis</strong> <strong>of</strong> <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong>. Evidence Report/Technology Assessment No. 43<br />
(Prepared by the University <strong>of</strong> California at San Francisco–Stanford Evidence-based Practice<br />
Center under Contract No. 290-97-0013), AHRQ <strong>Public</strong>ation No. 01-E058, Rockville, MD:<br />
Agency for Healthcare Research <strong>and</strong> Quality. July 2001.<br />
vi
CONTENTS<br />
Summary.........................................................................................................................1<br />
Evidence Report .............................................................................................................9<br />
PART I. OVERVIEW ..........................................................................................................11<br />
Chapter 1. An Introduction to the Report..............................................................................13<br />
1.1. General Overview .......................................................................................................13<br />
1.2. How to Use this Report ...............................................................................................18<br />
1.3. Acknowledgments.......................................................................................................20<br />
Chapter 2. Drawing on <strong>Safety</strong> <strong>Practices</strong> from Outside Health Care......................................25<br />
Chapter 3. Evidence-based Review Methodology ................................................................29<br />
PART II. REPORTING AND RESPONDING TO PATIENT SAFETY PROBLEMS ...........................39<br />
Chapter 4. Incident Reporting ...............................................................................................41<br />
Chapter 5. Root Cause <strong>Analysis</strong>............................................................................................51<br />
PART III. PATIENT SAFETY PRACTICES & TARGETS ...........................................................57<br />
Section A. Adverse Drug Events (ADEs).................................................................................58<br />
Chapter 6. Computerized Physician Order Entry (CPOE) with Clinical Decision Support<br />
Systems (CDSSs)....................................................................................................................59<br />
Chapter 7. The Clinical Pharmacist’s Role in Preventing Adverse Drug Events..................71<br />
Chapter 8. Computer Adverse Drug Event (ADE) Detection <strong>and</strong> Alerts..............................79<br />
Chapter 9. Protocols for High-Risk Drugs: Reducing Adverse Drug Events Related to<br />
Anticoagulants........................................................................................................................87<br />
Chapter 10. Unit-Dose Drug Distribution Systems .............................................................101<br />
Chapter 11. Automated Medication Dispensing Devices....................................................111<br />
Section B. Infection Control ...................................................................................................118<br />
Chapter 12. <strong>Practices</strong> to Improve H<strong>and</strong>washing Compliance .............................................119<br />
Chapter 13. Impact <strong>of</strong> Barrier Precautions in Reducing the Transmission <strong>of</strong> Serious<br />
Nosocomial Infections..........................................................................................................127<br />
Chapter 14. Impact <strong>of</strong> Changes in Antibiotic Use <strong>Practices</strong> on Nosocomial Infections <strong>and</strong><br />
Antimicrobial Resistance – Clostridium Difficile <strong>and</strong> Vancomycin-resistant Enterococcus<br />
(VRE)....................................................................................................................................141<br />
Chapter 15. Prevention <strong>of</strong> Nosocomial Urinary Tract Infections........................................149<br />
Subchapter 15.1. Use <strong>of</strong> Silver Alloy Urinary Catheters .................................................149<br />
Subchapter 15.2. Use <strong>of</strong> Suprapubic Catheters................................................................155<br />
Chapter 16. Prevention <strong>of</strong> Intravascular Catheter-Associated Infections............................163<br />
Subchapter 16.1. Use <strong>of</strong> Maximum Barrier Precautions During Central Venous Catheter<br />
Insertion.............................................................................................................................165<br />
Subchapter 16.2. Use <strong>of</strong> Central Venous Catheters Coated with Antibacterial or Antiseptic<br />
Agents ...............................................................................................................................169<br />
Subchapter 16.3. Use <strong>of</strong> Chlorhexidine Gluconate at the Central Venous Catheter<br />
Insertion Site .....................................................................................................................175<br />
Subchapter 16.4. Other <strong>Practices</strong> .....................................................................................178<br />
vii
Chapter 17. Prevention <strong>of</strong> Ventilator-Associated Pneumonia.............................................185<br />
Subchapter 17.1. <strong>Patient</strong> Positioning: Semi-recumbent Positioning <strong>and</strong> Continuous<br />
Oscillation .........................................................................................................................185<br />
Subchapter 17.2. Continuous Aspiration <strong>of</strong> Subglottic Secretions ..................................190<br />
Subchapter 17.3. Selective Digestive Tract Decontamination.........................................193<br />
Subchapter 17.4. Sucralfate <strong>and</strong> Prevention <strong>of</strong> VAP .......................................................198<br />
Section C. Surgery, Anesthesia, <strong>and</strong> Perioperative Medicine ................................................204<br />
Chapter 18. Localizing Care to High-Volume Centers .......................................................205<br />
Chapter 19. Learning Curves for New Procedures – the Case <strong>of</strong> Laparoscopic<br />
Cholecystectomy ..................................................................................................................213<br />
Chapter 20. Prevention <strong>of</strong> Surgical Site Infections ..............................................................221<br />
Subchapter 20.1. Prophylactic Antibiotics........................................................................221<br />
Subchapter 20.2. Perioperative Normothermia................................................................231<br />
Subchapter 20.3. Supplemental Perioperative Oxygen.....................................................234<br />
Subchapter 20.4. Perioperative Glucose Control ..............................................................237<br />
Chapter 21. Ultrasound Guidance <strong>of</strong> Central Vein Catheterization ....................................245<br />
Chapter 22. The Retained Surgical Sponge.........................................................................255<br />
Chapter 23. Pre-Anesthesia Checklists To Improve <strong>Patient</strong> <strong>Safety</strong> ....................................259<br />
Chapter 24. The Impact Of Intraoperative Monitoring On <strong>Patient</strong> <strong>Safety</strong> ..........................265<br />
Chapter 25. Beta-blockers <strong>and</strong> Reduction <strong>of</strong> Perioperative Cardiac Events .......................271<br />
Section D. <strong>Safety</strong> <strong>Practices</strong> for Hospitalized or Institutionalized Elders................................280<br />
Chapter 26. Prevention <strong>of</strong> Falls in Hospitalized <strong>and</strong> Institutionalized Older People........281<br />
Subchapter 26.1. Identification Bracelets for High-Risk <strong>Patient</strong>s ...................................284<br />
Subchapter 26.2. Interventions that Decrease the Use <strong>of</strong> Physical Restraints .................287<br />
Subchapter 26.3. Bed Alarms...........................................................................................291<br />
Subchapter 26.4. Special Hospital Flooring Materials to Reduce Injuries from <strong>Patient</strong> Falls<br />
...........................................................................................................................................293<br />
Subchapter 26.5. Hip Protectors to Prevent Hip Fracture ................................................295<br />
Chapter 27. Prevention <strong>of</strong> Pressure Ulcers in Older <strong>Patient</strong>s ..............................................301<br />
Chapter 28. Prevention <strong>of</strong> Delirium in Older Hospitalized <strong>Patient</strong>s ...................................307<br />
Chapter 29. Multidisciplinary Geriatric Consultation Services...........................................313<br />
Chapter 30. Geriatric Evaluation <strong>and</strong> Management Units for Hospitalized <strong>Patient</strong>s ...........323<br />
Section E. General Clinical Topics.........................................................................................331<br />
Chapter 31. Prevention <strong>of</strong> Venous Thromboembolism .......................................................333<br />
Chapter 32. Prevention <strong>of</strong> Contrast-Induced Nephropathy .................................................349<br />
Chapter 33. Nutritional Support ..........................................................................................359<br />
Chapter 34. Prevention <strong>of</strong> Clinically Significant Gastrointestinal Bleeding in Intensive Care<br />
Unit <strong>Patient</strong>s .........................................................................................................................368<br />
Chapter 35. Reducing Errors in the Interpretation <strong>of</strong> Plain Radiographs <strong>and</strong> Computed<br />
Tomography Scans ...............................................................................................................376<br />
Chapter 36. Pneumococcal Vaccination Prior to Hospital Discharge .................................385<br />
36.1. Vaccine Effectiveness .............................................................................................386<br />
36.2. Vaccine Delivery Methods......................................................................................388<br />
viii
Chapter 37. Pain Management.............................................................................................396<br />
Subchapter 37.1. Use <strong>of</strong> Analgesics in the Acute Abdomen ...........................................396<br />
Subchapter 37.2. Acute Pain Services..............................................................................400<br />
Subchapter 37.3. Prophylactic Antiemetics During <strong>Patient</strong>-controlled Analgesia Therapy<br />
...........................................................................................................................................404<br />
Subchapter 37.4. Non-pharmacologic Interventions for Postoperative Plan ...................406<br />
Section F. Organization, Structure, <strong>and</strong> Culture ....................................................................411<br />
Chapter 38. “Closed” Intensive Care Units <strong>and</strong> Other Models <strong>of</strong> Care for <strong>Critical</strong>ly Ill<br />
<strong>Patient</strong>s..................................................................................................................................413<br />
Chapter 39. Nurse Staffing, Models <strong>of</strong> Care Delivery, <strong>and</strong> Interventions ..........................423<br />
Chapter 40. Promoting a Culture <strong>of</strong> <strong>Safety</strong> ..........................................................................447<br />
Section G. Systems Issues <strong>and</strong> Human Factors ......................................................................458<br />
Chapter 41. Human Factors <strong>and</strong> <strong>Medical</strong> Devices ..............................................................459<br />
Subchapter 41.1. The Use <strong>of</strong> Human Factors in Reducing Device-related <strong>Medical</strong> Errors<br />
...........................................................................................................................................459<br />
Subchapter 41.2. Refining the Performance <strong>of</strong> <strong>Medical</strong> Device Alarms .........................462<br />
Subchapter 41.3. Equipment Checklists in Anesthesia....................................................466<br />
Chapter 42. Information Transfer.........................................................................................471<br />
Subchapter 42.1. Information Transfer Between Inpatient <strong>and</strong> Outpatient Pharmacies..471<br />
Subchapter 42.2. Sign-Out Systems for Cross-Coverage ................................................476<br />
Subchapter 42.3. Discharge Summaries <strong>and</strong> Follow-up ..................................................479<br />
Subchapter 42.4. Notifying <strong>Patient</strong>s <strong>of</strong> Abnormal Results ..............................................482<br />
Chapter 43. Prevention <strong>of</strong> Misidentifications......................................................................487<br />
Subchapter 43.1. Bar Coding ...........................................................................................487<br />
Subchapter 43.2. Strategies to Avoid Wrong-Site Surgery..............................................494<br />
Chapter 44. Crew Resource Management <strong>and</strong> its Applications in Medicine......................501<br />
Chapter 45. Simulator-Based Training <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong>..................................................511<br />
Chapter 46. Fatigue, Sleepiness, <strong>and</strong> <strong>Medical</strong> Errors..........................................................519<br />
Chapter 47. <strong>Safety</strong> During Transport <strong>of</strong> <strong>Critical</strong>ly Ill <strong>Patient</strong>s............................................534<br />
Subchapter 47.1. Interhospital Transport ..........................................................................536<br />
Subchapter 47.2. Intrahospital Transport .........................................................................538<br />
Section H. Role <strong>of</strong> the <strong>Patient</strong>.................................................................................................544<br />
Chapter 48. Procedures For Obtaining Informed Consent ..................................................546<br />
Chapter 49. Advance Planning For End-<strong>of</strong>-Life Care.........................................................556<br />
Chapter 50. Other <strong>Practices</strong> Related to <strong>Patient</strong> Participation ..............................................569<br />
PART IV. PROMOTING AND IMPLEMENTING SAFETY PRACTICES .......................................573<br />
Chapter 51. Practice Guidelines ..........................................................................................575<br />
Chapter 52. <strong>Critical</strong> Pathways .............................................................................................581<br />
Chapter 53. Clinical Decision Support Systems..................................................................589<br />
Chapter 54. Educational Techniques Used in Changing Provider Behavior.......................595<br />
Chapter 55. Legislation, Accreditation, <strong>and</strong> Market-Driven <strong>and</strong> Other Approaches to<br />
Improving <strong>Patient</strong> <strong>Safety</strong>......................................................................................................601<br />
ix
PART V. ANALYZING THE PRACTICES .............................................................................611<br />
Chapter 56. Methodology for Summarizing the Evidence for the <strong>Practices</strong> .......................613<br />
Chapter 57. <strong>Practices</strong> Rated by Strength <strong>of</strong> Evidence .........................................................619<br />
Chapter 58. <strong>Practices</strong> Rated by Research Priority...............................................................627<br />
Chapter 59. Listing <strong>of</strong> All <strong>Practices</strong>, Categorical Ratings, <strong>and</strong> Comments........................633<br />
Appendix: List <strong>of</strong> Contributors .................................................................................649<br />
Index............................................................................................................................655<br />
x
Summary<br />
Overview<br />
<strong>Patient</strong> safety has become a major concern <strong>of</strong> the general public <strong>and</strong> <strong>of</strong> policymakers at<br />
the State <strong>and</strong> Federal levels. This interest has been fueled, in part, by news coverage <strong>of</strong><br />
individuals who were the victims <strong>of</strong> serious medical errors <strong>and</strong> by the publication in 1999 <strong>of</strong> the<br />
Institute <strong>of</strong> Medicine’s (IOM’s) report To Err is Human: Building a Safer Health System. In its<br />
report, IOM highlighted the risks <strong>of</strong> medical care in the United States <strong>and</strong> shocked the<br />
sensibilities <strong>of</strong> many Americans, in large part through its estimates <strong>of</strong> the magnitude <strong>of</strong> medicalerrors-related<br />
deaths (44,000 to 98,000 deaths per year) <strong>and</strong> other serious adverse events. The<br />
report prompted a number <strong>of</strong> legislative <strong>and</strong> regulatory initiatives designed to document errors<br />
<strong>and</strong> begin the search for solutions. But Americans, who now wondered whether their next<br />
doctor’s or hospital visit might harm rather than help them, began to dem<strong>and</strong> concerted action.<br />
Three months after publication <strong>of</strong> the IOM report, an interagency Federal government<br />
group, the Quality Interagency Coordination Task Force (QuIC), released its response, Doing<br />
What Counts for <strong>Patient</strong> <strong>Safety</strong>: Federal Actions to Reduce <strong>Medical</strong> Errors <strong>and</strong> Their Impact.<br />
That report, prepared at the President’s request, both inventoried on-going Federal actions to<br />
reduce medical errors <strong>and</strong> listed more than 100 action items to be undertaken by Federal<br />
agencies.<br />
An action promised by the Agency for Healthcare Research <strong>and</strong> Quality (AHRQ), the<br />
Federal agency leading efforts to research <strong>and</strong> promote patient safety, was “the development <strong>and</strong><br />
dissemination <strong>of</strong> evidence-based, best safety practices to provider organizations.” To initiate the<br />
work to be done in fulfilling this promise, AHRQ commissioned the University <strong>of</strong> California at<br />
San Francisco (UCSF) – Stanford University Evidence-based Practice Center (EPC) in January<br />
2001 to review the scientific literature regarding safety improvement. To accomplish this, the<br />
EPC established an Editorial Board that oversaw development <strong>of</strong> this report by teams <strong>of</strong> content<br />
experts who served as authors.<br />
Defining <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong><br />
Working closely with AHRQ <strong>and</strong> the National Forum for Quality Measurement <strong>and</strong><br />
Reporting (the National Quality Forum, or NQF)—a public–private partnership formed in 1999<br />
to promote a national health care quality agenda the EPC began its work by defining a patient<br />
safety practice as a type <strong>of</strong> process or structure whose application reduces the probability <strong>of</strong><br />
adverse events resulting from exposure to the health care system across a range <strong>of</strong> diseases <strong>and</strong><br />
procedures.<br />
This definition is consistent with the dominant conceptual framework in patient safety,<br />
which holds that systemic change will be far more productive in reducing medical errors than<br />
will targeting <strong>and</strong> punishing individual providers. The definition’s focus on actions that cut<br />
across diseases <strong>and</strong> procedures also allowed the research team to distinguish patient safety<br />
activities from the more targeted quality improvement practices (e.g., practices designed to<br />
increase the use <strong>of</strong> beta-blockers in patients who are admitted to the hospital after having a<br />
myocardial infarction). The editors recognize, however, that this distinction is imprecise.<br />
1
This evidence-based review also focuses on hospital care as a starting point because the<br />
risks associated with hospitalization are significant, the strategies for improvement are better<br />
documented there than in other health care settings, <strong>and</strong> the importance <strong>of</strong> patient trust is<br />
paramount. The report, however, also considers evidence regarding other sites <strong>of</strong> care, such as<br />
nursing homes, ambulatory care, <strong>and</strong> patient self-management.<br />
The results <strong>of</strong> this EPC study will be used by the NQF to identify a set <strong>of</strong> proven patient<br />
safety practices that should be used by hospitals. Identification <strong>of</strong> these practices by NQF will<br />
allow patients throughout the nation to evaluate the actions their hospitals <strong>and</strong>/or health care<br />
facilities have taken to improve safety.<br />
Reporting The Evidence<br />
As is typical for evidence-based reviews, the goal was to provide a critical appraisal <strong>of</strong><br />
the evidence on the topic. This information would then be available to others to ensure that no<br />
practice unsupported by evidence would be endorsed <strong>and</strong> that no practice substantiated by a high<br />
level <strong>of</strong> pro<strong>of</strong> would lack endorsement. Readers familiar with the state <strong>of</strong> the evidence regarding<br />
quality improvement in areas <strong>of</strong> health care where this has been a research priority (e.g.,<br />
cardiovascular care) may be surprised <strong>and</strong> even disappointed, by the paucity <strong>of</strong> high quality<br />
evidence in other areas <strong>of</strong> health care for many patient safety practices. One reason for this is<br />
the relative youth <strong>of</strong> the field. Just as there had been little public recognition <strong>of</strong> the risks <strong>of</strong><br />
health care prior to the first IOM report, there has been relatively little attention paid to such<br />
risks – <strong>and</strong> strategies to mitigate them – among health pr<strong>of</strong>essionals <strong>and</strong> researchers.<br />
Moreover, there are a number <strong>of</strong> methodologic reasons why research in patient safety is<br />
particularly challenging. Many practices (e.g., the presence <strong>of</strong> computerized physician order<br />
entry systems, modifying nurse staffing levels) cannot be the subject <strong>of</strong> double-blind studies<br />
because their use is evident to the participants. Second, capturing all relevant outcomes,<br />
including “near misses” (such as a nurse catching an excessive dosage <strong>of</strong> a drug just before it is<br />
administered to a patient) <strong>and</strong> actual harm, is <strong>of</strong>ten very difficult. Third, many effective practices<br />
are multidimensional, <strong>and</strong> sorting out precisely which part <strong>of</strong> the intervention works is <strong>of</strong>ten<br />
quite challenging. Fourth, many <strong>of</strong> the patient safety problems that generate the most concern<br />
(wrong-site surgery, for example) are uncommon enough that demonstrating the success <strong>of</strong> a<br />
“safety practice” in a statistically meaningful manner with respect to outcomes is all but<br />
impossible.<br />
Finally, establishing firm epidemiologic links between presumed (<strong>and</strong> accepted) causes<br />
<strong>and</strong> adverse events is critical, <strong>and</strong> frequently difficult. For instance, in studying an intuitively<br />
plausible “risk factor” for errors, such as “fatigue,” analyses <strong>of</strong> errors commonly reveal the<br />
presence <strong>of</strong> fatigued providers (because many health care providers work long hours <strong>and</strong>/or late<br />
at night). The question is whether or not fatigue is over-represented among situations that lead<br />
to errors. The point is not that the problem <strong>of</strong> long work-hours should be ignored, but rather that<br />
strong epidemiologic methods need to be applied before concluding that an intuitive cause <strong>of</strong><br />
errors is, in fact, causal.<br />
Researchers now believe that most medical errors cannot be prevented by perfecting the<br />
technical work <strong>of</strong> individual doctors, nurses, or pharmacists. Improving patient safety <strong>of</strong>ten<br />
involves the coordinated efforts <strong>of</strong> multiple members <strong>of</strong> the health care team, who may adopt<br />
strategies from outside health care. The report reviews several practices whose evidence came<br />
2
from the domains <strong>of</strong> commercial aviation, nuclear safety, <strong>and</strong> aerospace, <strong>and</strong> the disciplines <strong>of</strong><br />
human factors engineering <strong>and</strong> organizational theory. Such practices include root cause analysis,<br />
computerized physician order entry <strong>and</strong> decision support, automated medication dispensing<br />
systems, bar coding technology, aviation-style preoperative checklists, promoting a “culture <strong>of</strong><br />
safety,” crew resource management, the use <strong>of</strong> simulators in training, <strong>and</strong> integrating human<br />
factors theory into the design <strong>of</strong> medical devices <strong>and</strong> alarms. In reviewing these practices, the<br />
research team sought to be flexible regarding st<strong>and</strong>ards <strong>of</strong> evidence, <strong>and</strong> included research<br />
evidence that would not have been considered for medical interventions. For example, the<br />
r<strong>and</strong>omized trial that is appropriately hailed as the “gold st<strong>and</strong>ard” in clinical medicine is not<br />
used in aviation, as this design would not capture all relevant information. Instead, detailed case<br />
studies <strong>and</strong> industrial engineering research approaches are utilized.<br />
Methodology<br />
To facilitate identification <strong>and</strong> evaluation <strong>of</strong> potential patient safety practices, the<br />
Editorial Board divided the content for the project into different domains. Some cover “content<br />
areas,” including traditional clinical areas such as adverse drug events, nosocomial infections,<br />
<strong>and</strong> complications <strong>of</strong> surgery, but also less traditional areas such as fatigue <strong>and</strong> information<br />
transfer. Other domains consist <strong>of</strong> practices drawn from broad (primarily nonmedical)<br />
disciplines likely to contain promising approaches to improving patient safety (e.g., information<br />
technology, human factors research, organizational theory). Once this list was created—with<br />
significant input from patient safety experts, clinician–researchers, AHRQ, <strong>and</strong> the NQF Safe<br />
<strong>Practices</strong> Committee—the editors selected teams <strong>of</strong> authors with expertise in the relevant subject<br />
matter <strong>and</strong>/or familiarity with the techniques <strong>of</strong> evidence-based review <strong>and</strong> technology appraisal.<br />
The authors were given explicit instructions regarding search strategies for identifying<br />
safety practices for evaluation (including explicit inclusion <strong>and</strong> exclusion criteria) <strong>and</strong> criteria for<br />
assessing each practice’s level <strong>of</strong> evidence for efficacy or effectiveness in terms <strong>of</strong> study design<br />
<strong>and</strong> study outcomes. Some safety practices did not meet the inclusion criteria because <strong>of</strong> the<br />
paucity <strong>of</strong> evidence regarding efficacy or effectiveness but were included in the report because<br />
an informed reader might reasonably expect them to be evaluated or because <strong>of</strong> the depth <strong>of</strong><br />
public <strong>and</strong> pr<strong>of</strong>essional interest in them. For such high pr<strong>of</strong>ile topics (such as bar coding to<br />
prevent misidentifications), the researchers tried to fairly present the practice’s background, the<br />
experience with the practice thus far, <strong>and</strong> the evidence (<strong>and</strong> gaps in the evidence) regarding the<br />
practice’s value.<br />
For each practice, authors were instructed to research the literature for information on:<br />
• prevalence <strong>of</strong> the problem targeted by the practice;<br />
• severity <strong>of</strong> the problem targeted by the practice;<br />
• the current utilization <strong>of</strong> the practice;<br />
• evidence on efficacy <strong>and</strong>/or effectiveness <strong>of</strong> the practice;<br />
• the practice’s potential for harm;<br />
• data on cost, if available; <strong>and</strong><br />
• implementation issues.<br />
The report presents the salient elements <strong>of</strong> each included study (e.g., study design,<br />
population/setting, intervention details, results), <strong>and</strong> highlights any important weaknesses <strong>and</strong><br />
3
iases <strong>of</strong> these studies. Authors were not asked to formally synthesize or combine the evidence<br />
across studies (e.g., perform a meta-analysis) as part <strong>of</strong> their task.<br />
The Editorial Board <strong>and</strong> the Advisory Panel reviewed the list <strong>of</strong> domains <strong>and</strong> practices to<br />
identify gaps in coverage. Submitted chapters were reviewed by the Editorial Board <strong>and</strong> revised<br />
by the authors, aided by feedback from the Advisory Panel. Once the content was finalized, the<br />
editors analyzed <strong>and</strong> ranked the practices using a methodology summarized below.<br />
Summarizing the Evidence <strong>and</strong> Rating the <strong>Practices</strong><br />
Because the report is essentially an anthology <strong>of</strong> a diverse <strong>and</strong> extensive group <strong>of</strong> patient<br />
safety practices with highly variable relevant evidence, synthesizing the findings was<br />
challenging, but necessary to help readers use the information. Two <strong>of</strong> the most obvious uses for<br />
this report are: 1) to inform efforts <strong>of</strong> providers <strong>and</strong> health care organizations to improve the<br />
safety <strong>of</strong> the care they provide, <strong>and</strong> 2) to inform AHRQ, other research agencies, <strong>and</strong><br />
foundations about potential fruitful investments for their research support. Other uses <strong>of</strong> the<br />
information are likely. In fact, the National Quality Forum plans to use this report to help<br />
identify a list <strong>of</strong> patient safety practices that consumers <strong>and</strong> others should know about as they<br />
choose among the health care provider organizations to which they have access.<br />
In an effort to assist both health care organizations interested in taking substantive<br />
actions to improve patient safety <strong>and</strong> research funders seeking to spend scarce resources wisely,<br />
AHRQ asked the EPC to rate the evidence <strong>and</strong> rank the practices by opportunity for safety<br />
improvement <strong>and</strong> by research priority. This report, therefore, contains two lists.<br />
To create these lists, the editors aimed to separate the practices that are most promising or<br />
effective from those that are least so on a range <strong>of</strong> dimensions, without implying any ability to<br />
calibrate a finely gradated scale for those practices in between. The editors also sought to<br />
present the ratings in an organized, accessible way while highlighting the limitations inherent in<br />
their rating schema. Proper metrics for more precise comparisons (e.g., cost-effectiveness<br />
analysis) require more data than are currently available in the literature.<br />
Three major categories <strong>of</strong> information were gathered to inform the rating exercise:<br />
• Potential Impact <strong>of</strong> the Practice: based on prevalence <strong>and</strong> severity <strong>of</strong> the patient<br />
safety target, <strong>and</strong> current utilization <strong>of</strong> the practice;<br />
• Strength <strong>of</strong> the Evidence Supporting the Practice: including an assessment <strong>of</strong> the<br />
relative weight <strong>of</strong> the evidence, effect size, <strong>and</strong> need for vigilance to reduce any<br />
potential negative collateral effects <strong>of</strong> the practice; <strong>and</strong><br />
• Implementation: considering costs, logistical barriers, <strong>and</strong> policy issues.<br />
For all <strong>of</strong> these data inputs into the practice ratings, the primary goal was to find the best<br />
available evidence from publications <strong>and</strong> other sources. Because the literature has not been<br />
previously organized with an eye toward addressing each <strong>of</strong> these areas, most <strong>of</strong> the estimates<br />
could be improved with further research, <strong>and</strong> some are informed by only general <strong>and</strong> somewhat<br />
speculative knowledge. In the summaries, the editors have attempted to highlight those<br />
assessments made with limited data.<br />
The four-person editorial team independently rated each <strong>of</strong> the 79 practices using general<br />
scores (e.g., High, Medium, Low) for a number <strong>of</strong> dimensions, including those italicized in the<br />
4
section above. The editorial team convened for 3 days in June, 2001 to compare scores, discuss<br />
disparities, <strong>and</strong> come to consensus about ratings for each category.<br />
In addition, each member <strong>of</strong> the team considered the totality <strong>of</strong> information on potential<br />
impact <strong>and</strong> support for a practice to score each <strong>of</strong> these factors on a 0 to 10 scale (creating a<br />
“Strength <strong>of</strong> the Evidence” list). For these ratings, the editors took the perspective <strong>of</strong> a leader <strong>of</strong><br />
a large health care enterprise (e.g., a hospital or integrated delivery system) <strong>and</strong> asked the<br />
question, “If I wanted to improve patient safety at my institution over the next 3 years <strong>and</strong><br />
resources were not a significant consideration, how would I grade this practice?” For this<br />
rating, the Editorial Board explicitly chose not to formally consider the difficulty or cost <strong>of</strong><br />
implementation in the rating. Rather, the rating simply reflected the strength <strong>of</strong> the evidence<br />
regarding the effectiveness <strong>of</strong> the practice <strong>and</strong> the probable impact <strong>of</strong> its implementation on<br />
reducing adverse events related to health care exposure. If the patient safety target was rated as<br />
“High” impact <strong>and</strong> there was compelling evidence (i.e., “High” relative study strength) that a<br />
particular practice could significantly reduce (e.g., “Robust” effect size) the negative<br />
consequences <strong>of</strong> exposure to the health care system (e.g., hospital-acquired infections), raters<br />
were likely to score the practice close to 10. If the studies were less convincing, the effect size<br />
was less robust, or there was a need for a “Medium” or “High” degree <strong>of</strong> vigilance because <strong>of</strong><br />
potential harms, then the rating would be lower.<br />
At the same time, the editors also rated the usefulness <strong>of</strong> conducting more research on<br />
each practice, emphasizing whether there appeared to be questions that a research program might<br />
have a reasonable chance <strong>of</strong> addressing successfully (creating a “Research Priority” list). Here,<br />
they asked themselves, “If I were the leader <strong>of</strong> a large agency or foundation committed to<br />
improving patient safety, <strong>and</strong> were considering allocating funds to promote additional research,<br />
how would I grade this practice?” If there was a simple gap in the evidence that could be<br />
addressed by a research study or if the practice was multifaceted <strong>and</strong> implementation could be<br />
eased by determining the specific elements that were effective, then the research priority was<br />
high. (For this reason, some practices are highly rated on both the “Strength <strong>of</strong> the Evidence”<br />
<strong>and</strong> “Research Priority” lists.) If the area was one <strong>of</strong> high potential impact (i.e., large number <strong>of</strong><br />
patients at risk for morbid or mortal adverse events) <strong>and</strong> a practice had been inadequately<br />
researched, then it would also receive a relatively high rating for research need. <strong>Practices</strong> might<br />
receive low research scores if they held little promise (e.g., relatively few patients are affected by<br />
the safety problem addressed by the practice or a significant body <strong>of</strong> knowledge already<br />
demonstrates the practice’s lack <strong>of</strong> utility). Conversely, a practice that was clearly effective, low<br />
cost, <strong>and</strong> easy to implement would not require further research <strong>and</strong> would also receive low<br />
research scores.<br />
In rating both the strength <strong>of</strong> the evidence <strong>and</strong> the research priority, the purpose was not<br />
to report precise 0 to 10 scores, but to develop general “zones” or practice groupings. This is<br />
important because better methods are available for making comparative ratings when the data<br />
inputs are available. The relative paucity <strong>of</strong> the evidence dissuaded the editors from using a<br />
more precise, sophisticated, but ultimately unfeasible, approach.<br />
5
Clear Opportunities for <strong>Safety</strong> Improvement<br />
The following 11 patient safety practices were the most highly rated (<strong>of</strong> the 79 practices<br />
reviewed in detail in the full report <strong>and</strong> ranked in the Executive Summary Addendum, AHRQ<br />
<strong>Public</strong>ation No. 01-E057b) in terms <strong>of</strong> strength <strong>of</strong> the evidence supporting more widespread<br />
implementation. <strong>Practices</strong> appear in descending order, with the most highly rated practices listed<br />
first. Because <strong>of</strong> the imprecision <strong>of</strong> the ratings, the editors did not further divide the practices,<br />
nor indicate where there were ties.<br />
• Appropriate use <strong>of</strong> prophylaxis to prevent venous thromboembolism in<br />
patients at risk;<br />
• Use <strong>of</strong> perioperative beta-blockers in appropriate patients to prevent<br />
perioperative morbidity <strong>and</strong> mortality;<br />
• Use <strong>of</strong> maximum sterile barriers while placing central intravenous catheters<br />
to prevent infections;<br />
• Appropriate use <strong>of</strong> antibiotic prophylaxis in surgical patients to prevent<br />
perioperative infections;<br />
• Asking that patients recall <strong>and</strong> restate what they have been told during the<br />
informed consent process;<br />
• Continuous aspiration <strong>of</strong> subglottic secretions (CASS) to prevent ventilatorassociated<br />
pneumonia;<br />
• Use <strong>of</strong> pressure relieving bedding materials to prevent pressure ulcers;<br />
• Use <strong>of</strong> real-time ultrasound guidance during central line insertion to prevent<br />
complications;<br />
• <strong>Patient</strong> self-management for warfarin (Coumadin) to achieve appropriate<br />
outpatient anticoagulation <strong>and</strong> prevent complications;<br />
• Appropriate provision <strong>of</strong> nutrition, with a particular emphasis on early enteral<br />
nutrition in critically ill <strong>and</strong> surgical patients; <strong>and</strong><br />
• Use <strong>of</strong> antibiotic-impregnated central venous catheters to prevent catheterrelated<br />
infections.<br />
This list is generally weighted toward clinical rather than organizational matters, <strong>and</strong><br />
toward care <strong>of</strong> the very, rather than the mildly or chronically ill. Although more than a dozen<br />
practices considered were general safety practices that have been the focus <strong>of</strong> patient safety<br />
experts for decades (i.e., computerized physician order entry, simulators, creating a “culture <strong>of</strong><br />
safety,” crew resource management), most research on patient safety has focused on more<br />
clinical areas. The potential application <strong>of</strong> practices drawn from outside health care has excited<br />
the patient safety community, <strong>and</strong> many such practices have apparent validity. However, clinical<br />
research has been promoted by the significant resources applied to it through Federal,<br />
foundation, <strong>and</strong> industry support. Since this study went where the evidence took it, more clinical<br />
practices rose to the top as potentially ready for implementation.<br />
6
Clear Opportunities for Research<br />
Until recently, patient safety research has had few champions, <strong>and</strong> even fewer champions<br />
with resources to bring to bear. The recent initiatives from AHRQ <strong>and</strong> other funders are a<br />
promising shift in this historical situation, <strong>and</strong> should yield important benefits.<br />
In terms <strong>of</strong> the research agenda for patient safety, the following 12 practices rated most<br />
highly, as follows:<br />
• Improved perioperative glucose control to decrease perioperative<br />
infections;<br />
• Localizing specific surgeries <strong>and</strong> procedures to high volume centers;<br />
• Use <strong>of</strong> supplemental perioperative oxygen to decrease perioperative<br />
infections;<br />
• Changes in nursing staffing to decrease overall hospital morbidity <strong>and</strong><br />
mortality;<br />
• Use <strong>of</strong> silver alloy-coated urinary catheters to prevent urinary tract<br />
infections;<br />
• Computerized physician order entry with computerized decision support<br />
systems to decrease medication errors <strong>and</strong> adverse events primarily due to<br />
the drug ordering process;<br />
• Limitations placed on antibiotic use to prevent hospital-acquired<br />
infections due to antibiotic-resistant organisms;<br />
• Appropriate use <strong>of</strong> antibiotic prophylaxis in surgical patients to prevent<br />
perioperative infections;<br />
• Appropriate use <strong>of</strong> prophylaxis to prevent venous thromboembolism in<br />
patients at risk;<br />
• Appropriate provision <strong>of</strong> nutrition, with a particular emphasis on early<br />
enteral nutrition in critically ill <strong>and</strong> post-surgical patients;<br />
• Use <strong>of</strong> analgesics in the patient with an acutely painful abdomen without<br />
compromising diagnostic accuracy; <strong>and</strong><br />
• Improved h<strong>and</strong>washing compliance (via education/behavior change; sink<br />
technology <strong>and</strong> placement; or the use <strong>of</strong> antimicrobial washing<br />
substances).<br />
Of course, the vast majority <strong>of</strong> the 79 practices covered in this report would benefit from<br />
additional research. In particular, some practices with longst<strong>and</strong>ing success outside <strong>of</strong> medicine<br />
(e.g., promoting a culture <strong>of</strong> safety) deserve further analysis, but were not explicitly ranked due<br />
to their unique nature <strong>and</strong> the present weakness <strong>of</strong> the evidentiary base in the health care<br />
literature.<br />
7
Conclusions<br />
This report represents a first effort to approach the field <strong>of</strong> patient safety through the lens<br />
<strong>of</strong> evidence-based medicine. Just as To Err is Human sounded a national alarm regarding patient<br />
safety <strong>and</strong> catalyzed other important commentaries regarding this vital problem, this review<br />
seeks to plant a seed for future implementation <strong>and</strong> research by organizing <strong>and</strong> evaluating the<br />
relevant literature. Although all those involved tried hard to include all relevant practices <strong>and</strong> to<br />
review all pertinent evidence, inevitably some <strong>of</strong> both were missed. Moreover, the effort to<br />
grade <strong>and</strong> rank practices, many <strong>of</strong> which have only the beginnings <strong>of</strong> an evidentiary base, was<br />
admittedly ambitious <strong>and</strong> challenging. It is hoped that this report provides a template for future<br />
clinicians, researchers, <strong>and</strong> policy makers as they extend, <strong>and</strong> inevitably improve upon, this<br />
work.<br />
In the detailed reviews <strong>of</strong> the practices, the editors have tried to define (to the extent<br />
possible from the literature) the associated costs—financial, operational, <strong>and</strong> political. However,<br />
these considerations were not factored into the summary ratings, nor were judgments made<br />
regarding the appropriate expenditures to improve safety. Such judgments, which involve<br />
complex trade<strong>of</strong>fs between public dollars <strong>and</strong> private ones, <strong>and</strong> between saving lives by<br />
improving patient safety versus doing so by investing in other health care or non-health care<br />
practices, will obviously be critical. However, the public reaction to the IOM report, <strong>and</strong> the<br />
media <strong>and</strong> legislative responses that followed it, seem to indicate that Americans are highly<br />
concerned about the risks <strong>of</strong> medical errors <strong>and</strong> would welcome public <strong>and</strong> private investment to<br />
decrease them. It seems logical to infer that Americans value safety during a hospitalization just<br />
as highly as safety during a transcontinental flight.<br />
8
Evidence Report
PART I. OVERVIEW<br />
Chapter 1. An Introduction to the Report<br />
1.1. General Overview<br />
1.2. How to Use this Report<br />
1.3. Acknowledgments<br />
Chapter 2. Drawing on <strong>Safety</strong> <strong>Practices</strong> from Outside Health Care<br />
Chapter 3. Evidence-based Review Methodology<br />
11
Chapter 1. An Introduction to the Report<br />
1.1. General Overview<br />
The Institute <strong>of</strong> Medicine’s (IOM) report, To Err is Human: Building a Safer Health<br />
System, 1 highlighted the risks <strong>of</strong> medical care in the United States. Although its prose was<br />
measured <strong>and</strong> its examples familiar to many in the health pr<strong>of</strong>essions (for example, the studies<br />
estimating that up to 98,000 Americans die each year from preventable medical errors were a<br />
decade old), the report shocked the sensibilities <strong>of</strong> many Americans. More importantly, the<br />
report undermined the fundamental trust that many previously had in the health care system.<br />
The IOM report prompted a number <strong>of</strong> legislative <strong>and</strong> regulatory initiatives designed to<br />
document errors <strong>and</strong> begin the search for solutions. These initiatives were further catalyzed by a<br />
second IOM report entitled Crossing the Quality Chasm: A New Health System for the 21 st<br />
Century, 2 which highlighted safety as one <strong>of</strong> the fundamental aims <strong>of</strong> an effective system. But<br />
Americans, who now wondered whether their next health care encounter might harm rather than<br />
help them, began to dem<strong>and</strong> concerted action.<br />
Making Health Care Safer represents an effort to determine what it is we might do in an<br />
effort to improve the safety <strong>of</strong> patients. In January 2001, the Agency for Healthcare Research<br />
<strong>and</strong> Quality (AHRQ), the Federal agency taking the lead in studying <strong>and</strong> promoting patient<br />
safety, commissioned the UCSF-Stanford Evidence-based Practice Center (EPC) to review the<br />
literature as it pertained to improving patient safety. In turn, the UCSF-Stanford EPC engaged 40<br />
authors at 11 institutions around the United States to review more than 3000 pieces <strong>of</strong> literature<br />
regarding patient safety practices. Although AHRQ expected that this evidence-based review<br />
would have multiple audiences, the National Quality Forum (NQF)—a public-private partnership<br />
formed in the Clinton Administration to promote a national quality agenda—was particularly<br />
interested in the results as it began its task <strong>of</strong> recommending <strong>and</strong> implementing patient safety<br />
practices supported by the evidence.<br />
A Definition <strong>of</strong> “<strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong>”<br />
One <strong>of</strong> our first tasks was to define “patient safety practices” in a manner that would<br />
allow us <strong>and</strong> our reviewers to assess the relevant evidence. Given our task—producing a full<br />
report in less than six months—a complete review <strong>of</strong> all practices associated with improving<br />
health care quality was both impossible <strong>and</strong> <strong>of</strong>f-point. Working closely with AHRQ <strong>and</strong> NQF,<br />
we chose the following definition:<br />
A <strong>Patient</strong> <strong>Safety</strong> Practice is a type <strong>of</strong> process or structure whose application reduces<br />
the probability <strong>of</strong> adverse events resulting from exposure to the health care system<br />
across a range <strong>of</strong> diseases <strong>and</strong> procedures.<br />
A few elements <strong>of</strong> the definition deserve emphasis. First, our focus on processes <strong>and</strong><br />
structure allowed us to emphasize changing the system to make it safer rather than targeting <strong>and</strong><br />
removing individual “bad apples.” We recognize that when individuals repeatedly perform<br />
poorly <strong>and</strong> are unresponsive to education <strong>and</strong> remediation, action is necessary. Nevertheless,<br />
there is virtual unanimity among patient safety experts that a focus on systemic change will be<br />
far more productive than an emphasis on finding <strong>and</strong> punishing poor performers.<br />
Second, looking at crosscutting diseases <strong>and</strong> procedures allowed us to distinguish patient<br />
safety activities from more targeted quality improvement practices. Admittedly, this<br />
dichotomization is imprecise. All would agree that a practice that makes it less likely that a<br />
13
patient will receive the wrong medication or have the wrong limb amputated is a patient safety<br />
practice. Most would also agree that practices designed to increase the use <strong>of</strong> beta-blockers in<br />
patients admitted to the hospital after myocardial infarction or to improve the technical<br />
performance <strong>of</strong> hernia repair would be quality improvement strategies rather than patient safety<br />
practices. When there was a close call, we generally chose to be inclusive. For example, we<br />
included practices designed to increase the rate <strong>of</strong> appropriate prophylaxis against venous<br />
thromboembolism, the appropriateness <strong>of</strong> pain management, <strong>and</strong> the ascertainment <strong>of</strong> patient<br />
preferences regarding end-<strong>of</strong>-life care. We recognize that these practices blur the line somewhat<br />
between safety <strong>and</strong> quality, but we believe that they are reasonable examples <strong>of</strong> ways to address<br />
potential patient safety hazards.<br />
Third, we realized it would be impossible to review every potential safety practice <strong>and</strong><br />
recognized that some gaps in the evidence were inevitable, so at times we reviewed illustrative<br />
examples that might be broadly generalizable. For example:<br />
• Methods to avoid misread radiographs (Chapter 35); where the content could<br />
be relevant to analogous efforts to avoid misread electrocardiograms or<br />
laboratory studies<br />
• Decreasing the risk <strong>of</strong> dangerous drugs (Chapter 9), where the focus was on<br />
anticoagulants, but similar considerations might be relevant for chemotherapy<br />
<strong>and</strong> other high-risk drugs<br />
• Localizing care to specialized providers reviews geriatric units <strong>and</strong><br />
intensivists (Chapters 30 <strong>and</strong> 38), but similar evidence may be relevant for the<br />
rapidly growing field <strong>of</strong> hospitalists 3-5<br />
• The use <strong>of</strong> ultrasound guidance for central line placement (Chapter 21); the<br />
premise (decreasing the risk <strong>of</strong> an invasive procedure through radiologic<br />
localization) may also be relevant for ultrasound guidance while performing<br />
other challenging procedures, such as thoracentesis<br />
A Focus on Hospital Care<br />
Most <strong>of</strong> the literature regarding medical errors has been drawn from hospital care. 6-22 For<br />
example, the two seminal studies on medical error 22,23 from which the <strong>of</strong>t-cited extrapolations <strong>of</strong><br />
yearly deaths from medical error were derived, have highlighted the risks <strong>of</strong> inpatient care. We<br />
applaud recent studies examining the risks <strong>of</strong> errors in the ambulatory setting 24 but believe that<br />
the hospital is an appropriate initial focus for an evidence-based review because the risks<br />
associated with hospitalization are high, strategies for improvement are better documented, <strong>and</strong><br />
the importance <strong>of</strong> patient trust is paramount.<br />
That said, the reader will see that we allowed the evidence to take us to other sites <strong>of</strong><br />
care. For example, although much <strong>of</strong> the literature regarding the occurrence <strong>and</strong> prevention <strong>of</strong><br />
adverse drug events is hospital-based, more recent literature highlights outpatient issues <strong>and</strong> is<br />
included in this review. An example is the chapter on decreasing the risk <strong>of</strong> anticoagulant<br />
treatment (Chapter 9), in which two <strong>of</strong> the most promising practices involve outpatient<br />
anticoagulation clinics <strong>and</strong> patient self-monitoring at home. Similarly, strategies to prevent falls<br />
or pressure ulcers are relevant to nursing home patients as well as those in hospitals, <strong>and</strong> many<br />
studies that shed light on these issues come from the former setting.<br />
14
The Evidence<br />
Chapter 3 describes our strategy for evidence review. As in other evidence-based<br />
reviews, we set the bar high. One would not want to endorse a practice unsupported by evidence,<br />
nor withhold one substantiated by a high level <strong>of</strong> pro<strong>of</strong>. In the end, we aimed to identify<br />
practices whose supporting evidence was so robust that immediate widespread implementation<br />
would lead to major improvements in patient safety. Additionally, we hoped to identify several<br />
practices whose promise merited a considerable investment in additional research, but whose<br />
evidentiary base was insufficient for immediate endorsement. The results <strong>of</strong> this effort are<br />
summarized in Part V <strong>of</strong> the Report.<br />
Readers familiar with the state <strong>of</strong> the evidence regarding quality improvement in areas<br />
where this has been a research priority (eg, cardiovascular care) may be surprised <strong>and</strong> even<br />
disappointed by the paucity <strong>of</strong> high quality evidence for many patient safety practices. The field<br />
is young. Just as there had been little public recognition <strong>of</strong> the risks <strong>of</strong> health care prior to the<br />
first IOM report, there has been relatively little attention paid to such risks—<strong>and</strong> strategies to<br />
mitigate them—among health pr<strong>of</strong>essionals <strong>and</strong> researchers. Nevertheless, we found a number <strong>of</strong><br />
practices supported by high quality evidence for which widespread implementation would save<br />
many thous<strong>and</strong>s <strong>of</strong> lives.<br />
Moreover, there are important methodologic reasons why research in patient safety is<br />
particularly challenging. First is the problem <strong>of</strong> blinding. The physician who has begun to use a<br />
new computerized order entry system cannot be blinded to the intervention or its purpose.<br />
Second, it is sometimes difficult to measure important outcomes. As in aviation, enormous<br />
benefits can be reaped from analyzing “near misses” (with no ultimate harm to patients), 25,26 <strong>and</strong><br />
yet these outcomes cannot be reliably counted in the absence <strong>of</strong> potentially obtrusive, <strong>and</strong> <strong>of</strong>ten<br />
very expensive observation. Third, many effective practices are multidimensional, <strong>and</strong> sorting<br />
out precisely which part <strong>of</strong> the intervention works is <strong>of</strong>ten quite challenging. Fourth, many <strong>of</strong> the<br />
patient safety problems that generate the most concern (wrong-site surgery, for example) are<br />
probably uncommon. This makes demonstrating the success <strong>of</strong> a “safety practice” in a<br />
statistically meaningful manner with respect to outcomes all but impossible.<br />
Finally, establishing firm epidemiologic links between presumed (<strong>and</strong> accepted) causes<br />
<strong>and</strong> adverse events is critical, <strong>and</strong> frequently difficult. For instance, verbal orders from doctors to<br />
nurses are regarded as a cause <strong>of</strong> medication errors almost as matter <strong>of</strong> dogma, with many<br />
hospitals prohibiting or strongly discouraging this practice except in emergency situations. 27 Yet,<br />
the one study that we could identify that specifically <strong>and</strong> comprehensively addressed this issue 28<br />
actually reported fewer errors among verbal medication orders compared with written medication<br />
orders. A similar relationship might be found studying other intuitively plausible “risk factors”<br />
for errors, such as “fatigue.” Because many health care providers work long hours <strong>and</strong>/or late at<br />
night, analyses <strong>of</strong> errors will commonly reveal fatigued providers. The question is whether or not<br />
fatigue is over-represented among situations that lead to errors. As discussed in Chapter 46, the<br />
evidence supporting fatigue as a contributor to adverse events is surprisingly mixed. The point is<br />
not that the problem <strong>of</strong> long work-hours should be ignored, but rather that strong epidemiologic<br />
methods need to be applied before concluding that an intuitive cause <strong>of</strong> errors is in fact causal.<br />
These methodologic issues are further explored in Chapters 3 (methods for analyzing the<br />
individual practices) <strong>and</strong> 56 (methods for summarizing the overall evidence).<br />
Improving patient safety is a team effort, <strong>and</strong> the playbook is <strong>of</strong>ten drawn from fields<br />
outside <strong>of</strong> health care. Most medical errors cannot be prevented by perfecting the technical work<br />
<strong>of</strong> individual doctors, nurses or pharmacists. Improving patient safety <strong>of</strong>ten involves the<br />
15
coordinated efforts <strong>of</strong> multiple members <strong>of</strong> the health care team, who may adopt strategies from<br />
outside health care. Thus, our teams <strong>of</strong> authors <strong>and</strong> advisors included physicians, pharmacists,<br />
nurses, <strong>and</strong> experts from non-medical fields. The literature we reviewed was <strong>of</strong>ten drawn from<br />
journals, books, or Web sites that will not be on most doctors’ reading lists. We reviewed several<br />
promising practices whose evidence came from the domains <strong>of</strong> commercial aviation, nuclear<br />
safety, <strong>and</strong> aerospace, <strong>and</strong> the disciplines <strong>of</strong> human factors engineering <strong>and</strong> organizational<br />
theory. In reviewing these practices, we tried to be flexible regarding st<strong>and</strong>ards <strong>of</strong> evidence. For<br />
example, the r<strong>and</strong>omized trial that is appropriately hailed as the “gold st<strong>and</strong>ard” in health care is<br />
rarely used in aviation, which instead relies on analyses <strong>of</strong> detailed case studies <strong>and</strong> industrial<br />
engineering research approaches. (Examples <strong>and</strong> additional discussion <strong>of</strong> this issue can be found<br />
in Chapter 2.)<br />
We also limited our discussion to the existing practices, recognizing that future<br />
technology may make the ones we reviewed obsolete. For example, much <strong>of</strong> the struggle to find<br />
safe ways to administer warfarin (Chapter 9) would be rendered moot by the development <strong>of</strong> a<br />
much safer, but equally effective oral anticoagulant that did not require monitoring. Similarly,<br />
the evidence regarding changing the flooring <strong>of</strong> rooms to decrease falls (Subchapter 26.4)<br />
indicated that present options may decrease the harm from falls but actually increase their rate.<br />
Clearly, a better surface would make falls both less likely <strong>and</strong> less harmful. Such a surface has<br />
not yet been tested.<br />
Finally, we have tried to define (to the extent possible from the literature) the costs—<br />
financial, operational, <strong>and</strong> political—associated with the patient safety practices we considered.<br />
However, we have not made judgments regarding the appropriate expenditures to improve<br />
safety. These judgments, which involve complex trade<strong>of</strong>fs between public dollars <strong>and</strong> private<br />
ones, <strong>and</strong> between saving lives by improving patient safety versus doing so by investing in other<br />
health care or non-health care practices, will obviously be critical. However, the public reaction<br />
to the IOM report, <strong>and</strong> the media <strong>and</strong> legislative responses that followed it, seem to indicate that<br />
Americans are highly concerned about the risks <strong>of</strong> medical errors <strong>and</strong> would welcome public <strong>and</strong><br />
private investment to decrease them. It seems logical to infer that Americans value safety during<br />
a hospitalization just as highly as safety during a transcontinental flight.<br />
The Decision to Include <strong>and</strong> Exclude <strong>Practices</strong><br />
The patient safety/quality interface was only one <strong>of</strong> several areas that called for<br />
judgments regarding which practices to include or exclude from the Report. In general (<strong>and</strong> quite<br />
naturally for an evidence-based review), we excluded those practices for which we found little or<br />
no supporting evidence. However, we recognize that patient safety is <strong>of</strong> great public <strong>and</strong><br />
pr<strong>of</strong>essional interest, <strong>and</strong> that the informed reader might expect to find certain topics in such a<br />
review. Therefore, we included several areas notwithst<strong>and</strong>ing their relatively meager evidentiary<br />
base. For such high pr<strong>of</strong>ile topics (such as bar coding to prevent misidentifications, Subchapter<br />
43.1), we tried to fairly present the practice’s background, the experience with the practice thus<br />
far, <strong>and</strong> the evidence (<strong>and</strong> gaps in the evidence) regarding its value. In many <strong>of</strong> these cases, we<br />
end by encouraging additional study or demonstration projects designed to prove whether the<br />
practices live up to their promise.<br />
Conversely, another very different group <strong>of</strong> practices lacked evidence <strong>and</strong> were excluded<br />
from the review. These practices were characterized by their largely self-evident value (in<br />
epidemiologic terms, their “face validity”). For example, large r<strong>and</strong>omized studies <strong>of</strong> the<br />
removal <strong>of</strong> concentrated potassium chloride from patient care floors surely are not necessary in<br />
order to recommend this practice as a sensible way <strong>of</strong> preventing egregious errors that should<br />
16
never occur. Although some <strong>of</strong> these types <strong>of</strong> practices were not included in this “evidencebased”<br />
Report, the reader should not infer their exclusion as a lack <strong>of</strong> endorsement.<br />
A cautionary note is in order when considering such “obviously beneficial” practices.<br />
Even an apparently straightforward practice like “signing the site” to prevent surgery or<br />
amputation <strong>of</strong> the wrong body part may lead to unexpected opportunities for error. As mentioned<br />
in Subchapter 43.2, some surgeons adopt the practice <strong>of</strong> marking the intended site, while others<br />
mark the site to avoid. The clinical research literature furnishes enough examples <strong>of</strong> practices<br />
that everyone “knew” to be beneficial but proved not to be (or even proved to be harmful) once<br />
good studies were conducted (antiarrhythmic therapy for ventricular ectopy 29 or hormone<br />
replacement therapy to prevent cardiac deaths, 30 for example) that it is reasonable to ask for<br />
high-quality evidence for most practices. This is particularly true when practices are expensive,<br />
complex to implement, or carry their own risks.<br />
Other Content Issues<br />
There may appear to some readers to be an inordinate focus on clinical issues versus<br />
more general patient safety practices. In this <strong>and</strong> other matters, we went where the evidence took<br />
us. Although more than a dozen chapters <strong>of</strong> the Report consider general safety practices that<br />
have been the focus <strong>of</strong> many patient safety experts for decades (ie, computerized order entry,<br />
simulators, crew resource management), most research on patient safety, in fact, has focused on<br />
more clinical matters. It is likely that some <strong>of</strong> this is explained by the previous “disconnect”<br />
between research in patient safety <strong>and</strong> its application. We are hopeful that the Report helps to<br />
bridge this gap. We also think it likely that clinical research has been promoted by the significant<br />
resources applied to it through Federal, foundation, <strong>and</strong> industry support. Until recently, patient<br />
safety research has had few champions, <strong>and</strong> even fewer champions with resources. The recent<br />
initiatives from AHRQ <strong>and</strong> other funders are a promising shift in this historical situation, <strong>and</strong><br />
should yield important benefits.<br />
The reader will notice that there is relatively little specific coverage <strong>of</strong> issues in<br />
pediatrics, obstetrics, <strong>and</strong> psychiatry. Most <strong>of</strong> the patient safety practices we reviewed have<br />
broad applicability to those fields as well as larger fields such as surgery <strong>and</strong> medicine. Much <strong>of</strong><br />
the research in the former fields was too disease-specific to include in this volume. For example,<br />
practices to improve the safety <strong>of</strong> childbirth, although exceptionally important, were excluded<br />
because they focused on the care <strong>of</strong> patients with a single “condition,” just as we excluded<br />
research focused specifically on the care <strong>of</strong> patients with pneumonia or stroke.<br />
Readers may also be surprised by the relatively small portion <strong>of</strong> the Report devoted to the<br />
prevention <strong>of</strong> high-pr<strong>of</strong>ile <strong>and</strong> “newsworthy” errors. Even if much <strong>of</strong> the national attention to<br />
patient safety stemmed from concerns about wrong-site surgery or transfusion mix-ups, in fact<br />
these are not the dominant patient safety problems today. If widespread use <strong>of</strong> hip protectors<br />
(Subchapter 26.5) leads to a marked decrease in injuries from patient falls, implementing this<br />
safety practice would be more important than preventing the few wrong-site surgeries each year,<br />
although the former seem far less likely to garner attention in a tabloid.<br />
Conclusions<br />
Making Health Care Safer represents a first effort to approach the field <strong>of</strong> patient safety<br />
through the lens <strong>of</strong> evidence-based medicine. Just as To Err is Human sounded a national alarm<br />
regarding patient safety <strong>and</strong> catalyzed other important commentaries regarding this vital<br />
problem, this review is a germinal effort to mine the relevant literature. Although we <strong>and</strong> the<br />
authors tried hard to include all relevant practices <strong>and</strong> to review all pertinent evidence, we<br />
17
inevitably missed some <strong>of</strong> both. Moreover, our effort to rank practices (Part V), many <strong>of</strong> which<br />
have only the beginnings <strong>of</strong> an evidentiary base, was admittedly ambitious <strong>and</strong> challenging. We<br />
hope that the Report provides a template for future clinicians, researchers, <strong>and</strong> policy makers as<br />
they extend, <strong>and</strong> inevitably improve upon, our work.<br />
1.2. How to Use this Report<br />
Organizational Framework<br />
This document is divided into five parts:<br />
Part I – The overview introduces many <strong>of</strong> the methodologic, content, <strong>and</strong> policy issues.<br />
Part II – We describe, <strong>and</strong> present the evidence regarding 2 practices that are used to report <strong>and</strong><br />
respond to patient safety problems: incident reporting <strong>and</strong> root cause analysis. Since both these<br />
“practices” have relevance for all <strong>of</strong> the patient safety targets <strong>and</strong> practices covered in Part III,<br />
we neither grade them nor rank them.<br />
Part III – In 45 chapters, we review the evidence regarding the utility <strong>of</strong> 79 patient safety<br />
practices. Each chapter is structured in a st<strong>and</strong>ard fashion, as follows:<br />
• Background – <strong>of</strong> the patient safety problem <strong>and</strong> the practice;<br />
• Practice Description – in which we try to present the practice at a level <strong>of</strong><br />
detail that would allow a reader to determine the practice’s applicability to<br />
their setting;<br />
• Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem – Here, we try to<br />
answer the following questions: How common is the safety problem the<br />
practice is meant to address? How <strong>of</strong>ten does the problem lead to harm? How<br />
bad is the harm when it occurs?;<br />
• Opportunities for Impact – In this section, we consider the present-day use <strong>of</strong><br />
the patient safety practice. For example, we found that the use <strong>of</strong> “unit-dose”<br />
drug dispensing was quite common in US hospitals, <strong>and</strong> thus the opportunity<br />
to make an impact with wider dissemination <strong>of</strong> this practice was relatively<br />
low. Conversely, computerized physician order entry is still relatively<br />
uncommon, <strong>and</strong> therefore (assuming it is effective), its widespread<br />
implementation could have a far larger impact;<br />
• Study Designs – We review the designs <strong>of</strong> the major studies evaluating the<br />
practice. Similarly, Study Outcomes looks at the kinds <strong>of</strong> outcomes (eg,<br />
adverse drug events, surgical complications, mortality) that were considered.<br />
Our criteria for grading the evidence related to both design <strong>and</strong> outcomes<br />
(more information on other methodologic issues appears in Chapter 3);<br />
• Evidence for Effectiveness <strong>of</strong> the Practice – Here, the authors summarize the<br />
findings <strong>of</strong> the studies <strong>and</strong> comment on any methodologic concerns that might<br />
effect the strength <strong>of</strong> these findings. This section is <strong>of</strong>ten accompanied by<br />
tables summarizing the studies <strong>and</strong> their findings;<br />
18
• Potential for Harm – Many practices that are effective in improving patient<br />
safety nonetheless carry the potential for harm. More widespread use <strong>of</strong><br />
antibiotic prophylaxis or antibiotic-impregnated urinary or vascular catheters<br />
could prevent individual hospital-acquired infections yet breed antibiotic<br />
resistance. Increasing the use <strong>of</strong> barrier precautions could also prevent<br />
infections, but might lead caregivers to visit patients less <strong>of</strong>ten. These sections<br />
do not imply that harm is inevitable; rather they highlight the issues that<br />
require vigilance during the implementation <strong>of</strong> effective practices;<br />
• Costs <strong>and</strong> Implementation – Here we consider the costs <strong>and</strong> other challenges<br />
<strong>of</strong> implementing the practice. We tried to uncover data related to the true costs<br />
<strong>of</strong> implementation (How much does an automatic drug dispensing machine<br />
cost a pharmacy?), but also considered some <strong>of</strong> the potential <strong>of</strong>fsets when<br />
there were data available. We also considered issues <strong>of</strong> feasibility: How much<br />
behavior change would be necessary to implement the practice? Would there<br />
be major political concerns or important shifts in who pays for care or is<br />
compensated for providing it? We tried not to assign values to such issues, but<br />
rather to present them so that policy makers could consider them; <strong>and</strong><br />
• Comment – Here, the authors highlight the state <strong>of</strong> the evidence, elucidate key<br />
implementation issues, <strong>and</strong> define a potential research agenda.<br />
Part IV – In many ways a mirror <strong>of</strong> Part II, Part IV considers the ways in which patient safety<br />
practices can be implemented. The evidence is reviewed, <strong>and</strong> some <strong>of</strong> the benefits <strong>and</strong><br />
limitations <strong>of</strong> various strategies are analyzed. As with Part II, we neither grade nor rank these<br />
“practices” in Part V since each <strong>of</strong> these strategies can be applied to most <strong>of</strong> the patient safety<br />
targets <strong>and</strong> practices covered in Part III.<br />
Part V – Here we analyze the practices. Using methods described in Chapter 56, we synthesize<br />
the evidence in Part III to grade <strong>and</strong> rank the patient safety practices across two major<br />
dimensions:<br />
• Does the evidence support implementation <strong>of</strong> the practice to improve patient safety?<br />
• Does the evidence support additional research into the practice?<br />
Tips for Users <strong>of</strong> the Report<br />
We envision that this evidence-based report <strong>of</strong> patient safety practices will be useful to a<br />
wide audience.<br />
Policy makers may use its contents <strong>and</strong> recommendations to promote or fund the<br />
implementation <strong>of</strong> certain practices. Similarly, local health care organization leaders (including<br />
leaders <strong>of</strong> hospitals, medical groups, or integrated delivery systems) may use the data <strong>and</strong><br />
analysis to choose which practices to consider implementing or further promoting at their<br />
institutions.<br />
Researchers will identify a wealth <strong>of</strong> potential research opportunities. This document is,<br />
in many ways, a road map for future research into patient safety. Those who fund research,<br />
including (but not limited to) AHRQ, which sponsored this report, will find literally dozens <strong>of</strong><br />
areas ripe for future studies. In some cases, such studies may be expensive r<strong>and</strong>omized<br />
controlled trials, while other practices may require a simple meta-analysis or cost-effectiveness<br />
analysis to tip the scales toward or away from recommending a practice.<br />
19
Clinicians <strong>and</strong> trainees will, we hope, find the material both interesting <strong>and</strong> relevant to<br />
their practices. One <strong>of</strong> the salutary consequences <strong>of</strong> the IOM’s reports has been their impact on<br />
the attitudes <strong>of</strong> our future health care providers. We have noticed at our institutions that students<br />
<strong>and</strong> post-graduate trainees in medicine, nursing, <strong>and</strong> pharmacy are increasingly taking a systems<br />
approach to health care. Several <strong>of</strong> us have heard medical residents refer to issues as “patient<br />
safety problems” that beg for a “systems solution” over the past two years, terms that were<br />
absent from the medical ward a few years earlier. Clinicians must be part <strong>of</strong> the solutions to<br />
patient safety problems, <strong>and</strong> their increasing interest in the field is an exceedingly hopeful sign.<br />
Finally, although not primarily written for patients <strong>and</strong> their families, we recognize the<br />
broad public interest in, <strong>and</strong> concern about patient safety <strong>and</strong> believe that much <strong>of</strong> the material<br />
will be compelling <strong>and</strong> potentially useful to the public. For years quality advocates have<br />
lamented the relatively small impact that “quality report cards” appear to have on patients’<br />
choices <strong>of</strong> health care providers <strong>and</strong> institutions. One study demonstrated that patients were more<br />
likely to respond to a newspaper report <strong>of</strong> an egregious error than such quality report cards. 31<br />
These data indicate that patients may be interested in knowing whether their institutions,<br />
providers, <strong>and</strong> health plans are proactive in implementing practices that demonstrably decrease<br />
the risk <strong>of</strong> adverse events. Also, any general reader is likely to come away from this Report with<br />
heightened sensitivity to the unique challenges that the health care industry—which aims to<br />
provide compassionate, individualized care in a dynamic, organizationally <strong>and</strong> politically<br />
complex, <strong>and</strong> technologically fluid environment—faces in improving safety, <strong>and</strong> the significant<br />
strides that have already been made. Continued improvement will require the infusion <strong>of</strong><br />
substantial resources, <strong>and</strong> the public debate about their source, quantity, <strong>and</strong> target is likely to be<br />
lively <strong>and</strong> very important.<br />
1.3. Acknowledgments<br />
We are grateful to our authors, whose passion for the evidence overcame the tremendous<br />
time pressure driven by an unforgiving timeline. They succumbed to our endless queries, as<br />
together we attempted to link the literature from a variety <strong>of</strong> areas into an evidence-based<br />
repository whose presentation respected an overarching framework, notwithst<strong>and</strong>ing the variety<br />
<strong>of</strong> practices reviewed. Our Managing Editor, Amy J. Markowitz, JD, edited the manuscript<br />
with incredible skill <strong>and</strong> creativity. Each part was reviewed <strong>and</strong> improved multiple times because<br />
<strong>of</strong> her unending dedication to the project. Susan B. Nguyen devoted countless hours assisting<br />
the team in every imaginable task, <strong>and</strong> was critical in keeping the project organized <strong>and</strong> on track.<br />
Invaluable editorial <strong>and</strong> copyediting assistance was provided by Kathleen Kerr, Talia Baruth,<br />
<strong>and</strong> Mary Whitney. We thank Phil Tiso for helping produce the electronic version <strong>of</strong> the<br />
Report. We also thank A. Eugene Washington, MD, MSc, Director <strong>of</strong> the UCSF-Stanford<br />
Evidence-based Practice Center, for his guidance <strong>and</strong> support.<br />
We were aided by a blue-ribbon Advisory Panel whose suggestions <strong>of</strong> topics, resources,<br />
<strong>and</strong> ideas were immensely helpful to us throughout the life <strong>of</strong> the project. Given the time<br />
constraints, there were doubtless many constructive suggestions that we were unable to<br />
incorporate into the Report. The responsibility for such omissions is ours alone. We want to<br />
recognize the important contribution <strong>of</strong> our advisors to our thinking <strong>and</strong> work. The Advisory<br />
Panel’s members are:<br />
• David M. Gaba, MD, Director, <strong>Patient</strong> <strong>Safety</strong> Center <strong>of</strong> Inquiry at Veterans<br />
Affairs Palo Alto Health Care System, Pr<strong>of</strong>essor <strong>of</strong> Anesthesia, Stanford<br />
University School <strong>of</strong> Medicine<br />
20
• John W. Gosbee, MD, MS, Director <strong>of</strong> <strong>Patient</strong> <strong>Safety</strong> Information Systems,<br />
Department <strong>of</strong> Veterans Affairs National Center for <strong>Patient</strong> <strong>Safety</strong><br />
• Peter V. Lee, JD, President <strong>and</strong> Chief Executive Officer, Pacific Business<br />
Group on Health<br />
• Arnold Milstein, MD, MPH, <strong>Medical</strong> Director, Pacific Business Group on<br />
Health; National Health Care Thought Leader, William M. Mercer, Inc.<br />
• Karlene H. Roberts, PhD, Pr<strong>of</strong>essor, Walter A. Haas School <strong>of</strong> Business,<br />
University <strong>of</strong> California, Berkeley<br />
• Stephen M. Shortell, PhD, Blue Cross <strong>of</strong> California Distinguished Pr<strong>of</strong>essor<br />
<strong>of</strong> Health Policy <strong>and</strong> Management <strong>and</strong> Pr<strong>of</strong>essor <strong>of</strong> Organization Behavior,<br />
University <strong>of</strong> California, Berkeley<br />
We thank the members <strong>of</strong> the National Quality Forum’s (NQF) Safe <strong>Practices</strong><br />
Committee, whose co-chairs, Maureen Bisognano <strong>and</strong> Henri Manasse, Jr., PhD, ScD, <strong>and</strong><br />
members helped set our course at the outset <strong>of</strong> the project. NQF President <strong>and</strong> Chief Executive<br />
Officer Kenneth W. Kizer, MD, MPH, Vice President Elaine J. Power, <strong>and</strong> Program Director<br />
Laura N. Blum supported the project <strong>and</strong> coordinated their committee members’ helpful<br />
suggestions as the project proceeded. We appreciate their patience during the intensive research<br />
period as they waited for the final product.<br />
Finally, we are indebted to the Agency for Healthcare Research <strong>and</strong> Quality (AHRQ),<br />
which funded the project <strong>and</strong> provided us the intellectual support <strong>and</strong> resources needed to see it<br />
to successful completion. Jacqueline Besteman, JD <strong>of</strong> AHRQ’s Center for Practice <strong>and</strong><br />
Technology Assessment, <strong>and</strong> Gregg Meyer, MD, <strong>and</strong> Nancy Foster, both from AHRQ’s Center<br />
for Quality Improvement <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong>, were especially important to this project. John<br />
Eisenberg, MD, MBA, AHRQ’s Administrator, was both a tremendous supporter <strong>of</strong> our work<br />
<strong>and</strong> an inspiration for it.<br />
The Editors<br />
San Francisco <strong>and</strong> Palo Alto, California<br />
June, 2001<br />
References<br />
1. Kohn L, Corrigan J, Donaldson M, editors. To Err Is Human: Building a Safer Health<br />
System. Washington, DC: Committee on Quality <strong>of</strong> Health Care in America, Institute <strong>of</strong><br />
Medicine. National Academy Press; 2000.<br />
2. Committee on Quality <strong>of</strong> Health Care in America, Institute <strong>of</strong> Medicine, editor. Crossing<br />
the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National<br />
Academy Press; 2001.<br />
3. Wachter RM, Goldman L. The emerging role <strong>of</strong> "hospitalists" in the American health care<br />
system. N Engl J Med. 1996;335:514-517.<br />
4. Wachter RM, Katz P, Showstack J, Bindman AB, Goldman L. Reorganizing an academic<br />
medical service: impact on cost, quality, patient satisfaction, <strong>and</strong> education. JAMA.<br />
1998;279:1560-1565.<br />
21
5. Meltzer DO, Manning WG, Shah M, Morrison J, Jin L, Levinson W. Decreased length <strong>of</strong><br />
stay, costs <strong>and</strong> mortality in a r<strong>and</strong>omized trial <strong>of</strong> academic hospitalists. Paper presented at:<br />
Society <strong>of</strong> General Internal Medicine-24th Annual Meeting; May 2-5, 2001; San Diego,<br />
CA.<br />
6. Safren MA, Chapanis A. A critical incident study <strong>of</strong> hospital medication errors. Hospitals.<br />
1960;34:32-34.<br />
7. Mills DH, editor. Report on the medical insurance feasibility study / sponsored jointly by<br />
California <strong>Medical</strong> Association <strong>and</strong> California Hospital Association. San Francisco: Sutter<br />
<strong>Public</strong>ations, Inc.; 1977.<br />
8. Bosk CL. Forgive <strong>and</strong> remember : managing medical failure. Chicago: University <strong>of</strong><br />
Chicago Press; 1979.<br />
9. Abramson NS, Wald KS, Grenvik AN, Robinson D, Snyder JV. Adverse occurrences in<br />
intensive care units. JAMA. 1980;244:1582-1584.<br />
10. Barker KN, Mikeal RL, Pearson RE, Illig NA, Morse ML. Medication errors in nursing<br />
homes <strong>and</strong> small hospitals. Am J Hosp Pharm. 1982;39:987-991.<br />
11. Cooper JB, Newbower RS, Kitz RJ. An analysis <strong>of</strong> major errors <strong>and</strong> equipment failures in<br />
anesthesia management: considerations for prevention <strong>and</strong> detection. Anesthesiology.<br />
1984;60:34-42.<br />
12. Rubins HB, Moskowitz MA. Complications <strong>of</strong> care in a medical intensive care unit. J Gen<br />
Intern Med. 1990;5:104-109.<br />
13. Silber JH, Williams SV, Krakauer H, Schwartz JS. Hospital <strong>and</strong> patient characteristics<br />
associated with death after surgery. A study <strong>of</strong> adverse occurrence <strong>and</strong> failure to rescue.<br />
Med Care. 1992;30:615-629.<br />
14. Runciman WB, Sellen A, Webb RK, Williamson JA, Currie M, Morgan C, et al. The<br />
Australian Incident Monitoring Study. Errors, incidents <strong>and</strong> accidents in anaesthetic<br />
practice. Anaesth Intensive Care. 1993;21:506-519.<br />
15. Donchin Y, Gopher D, Olin M, Badihi Y, Biesky M, Sprung CL, et al. A look into the<br />
nature <strong>and</strong> causes <strong>of</strong> human errors in the intensive care unit. <strong>Critical</strong> Care Medicine.<br />
1995;23:294-300.<br />
16. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events. Implications for prevention. ADE<br />
Prevention Study Group. JAMA. 1995;274:29-34.<br />
17. Dean BS, Allan EL, Barber ND, Barker KN. Comparison <strong>of</strong> medication errors in an<br />
American <strong>and</strong> a British hospital. Am J Health Syst Pharm. 1995;52:2543-2549.<br />
18. Wilson RM, Runciman WB, Gibberd RW, Harrison BT, Newby L, Hamilton JD. The<br />
Quality in Australian Health Care Study. Med J Aust. 1995;163:458-471.<br />
19. Andrews LB, Stocking C, Krizek T, Gottlieb L, Krizek C, Vargish T, et al. An alternative<br />
strategy for studying adverse events in medical care. Lancet. 1997;349:309-313.<br />
20. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in<br />
hospitalized patients. Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA.<br />
1997;277:301-306.<br />
21. Lesar TS, Lomaestro BM, Pohl H. Medication-prescribing errors in a teaching hospital. A<br />
9-year experience. Arch Intern Med. 1997;157:1569-1576.<br />
22. Thomas EJ, Studdert DM, Burstin HR, Orav EJ, Zeena T, Williams EJ, et al. Incidence <strong>and</strong><br />
types <strong>of</strong> adverse events <strong>and</strong> negligent care in Utah <strong>and</strong> Colorado. Med Care. 2000;38:261-<br />
271.<br />
22
23. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, et al. Incidence<br />
<strong>of</strong> adverse events <strong>and</strong> negligence in hospitalized patients. Results <strong>of</strong> the Harvard <strong>Medical</strong><br />
Practice Study I. N Engl J Med. 1991;324:370-376.<br />
24. G<strong>and</strong>hi TK, Burstin HR, Cook EF, Puopolo AL, Haas JS, Brennan TA, et al. Drug<br />
complications in outpatients. J Gen Intern Med. 2000;15:149-154.<br />
25. Barach P, Small SD. Reporting <strong>and</strong> preventing medical mishaps: lessons from non-medical<br />
near miss reporting systems. BMJ. 2000;320:759-763.<br />
26. Leape LL. Reporting <strong>of</strong> medical errors: time for a reality check. Qual Health Care.<br />
2000;9:144-145.<br />
27. Dahl FC, Davis NM. A survey <strong>of</strong> hospital policies on verbal orders. Hosp Pharm.<br />
1990;25:443-447.<br />
28. West DW, Levine S, Magram G, MacCorkle AH, Thomas P, Upp K. Pediatric medication<br />
order error rates related to the mode <strong>of</strong> order transmission. Arch Pediatr Adolesc Med.<br />
1994;148:1322-1326.<br />
29. Preliminary report: effect <strong>of</strong> encainide <strong>and</strong> flecainide on mortality in a r<strong>and</strong>omized trial <strong>of</strong><br />
arrhythmia suppression after myocardial infarction. The Cardiac Arrhythmia Suppression<br />
Trial (CAST) Investigators. N Engl J Med. 1989;321:406-412.<br />
30. Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, et al. R<strong>and</strong>omized trial <strong>of</strong><br />
estrogen plus progestin for secondary prevention <strong>of</strong> coronary heart disease in<br />
postmenopausal women. Heart <strong>and</strong> Estrogen/progestin Replacement Study (HERS)<br />
Research Group. JAMA. 1998;280:605-613.<br />
31. Mennemeyer ST, Morrisey MA, Howard LZ. Death <strong>and</strong> reputation: how consumers acted<br />
upon HCFA mortality information. Inquiry. 1997;34:117-128.<br />
23
Chapter 2. Drawing on <strong>Safety</strong> <strong>Practices</strong> from Outside Health Care<br />
The medical pr<strong>of</strong>ession’s previous inattention to medical error, along with other<br />
publicized deficiencies (such as a notable lag in adopting sophisticated information technologies)<br />
have invited unfavorable comparisons between health care <strong>and</strong> other complex industries. 1-5 The<br />
first <strong>of</strong> the two recent Institute <strong>of</strong> Medicine (IOM) reports on the quality <strong>of</strong> health care in<br />
America, To Err is Human: Building a Safer Health System, 3 states that “health care is a decade<br />
or more behind other high-risk industries in its attention to ensuring basic safety.” Consequently,<br />
one <strong>of</strong> the goals <strong>of</strong> this project was to search these other industries for evidence-based safety<br />
strategies that might be applied to health care.<br />
The relatively short timeline <strong>of</strong> this project necessitated a focused approach to the search<br />
for potentially applicable patient safety practices from non-health care writings. Fortunately,<br />
many relevant practices have received at least some analysis or empirical study in the health care<br />
literature. As a practical solution we present original articles from outside health care as<br />
foundational <strong>and</strong> background material, rather than as a primary source <strong>of</strong> evidence. Specific<br />
topics <strong>and</strong> practices reviewed in Making Health Care Safer that clearly derive from fields<br />
outside health care include:<br />
• Incident reporting (Chapter 4)<br />
• Root cause analysis (Chapter 5)<br />
• Computerized physician order entry <strong>and</strong> decision support as a means <strong>of</strong><br />
reducing medication errors (Chapter 6)<br />
• Automated medication dispensing systems (Chapter 11)<br />
• Bar coding technology to avoid misidentification errors (Subchapter 43.1)<br />
• Aviation-style preoperative checklists for anesthesia equipment (Chapter 23<br />
<strong>and</strong> Subchapter 41.3)<br />
• Promoting a “culture <strong>of</strong> safety” (Chapter 40)<br />
• Crew resource management, a model for teamwork training <strong>and</strong> crisis<br />
response modeled after training approaches in aviation (Chapter 44)<br />
• Simulators (<strong>of</strong> patients or clinical scenarios) as a training tool (Chapter 45)<br />
• Human factors theory in the design <strong>of</strong> medical devices <strong>and</strong> alarms (Chapter<br />
41)<br />
Many readers may still wonder at the relative paucity <strong>of</strong> safety practices drawn from nonhealth<br />
care sources. While the headline-grabbing assessments <strong>of</strong> medicine’s safety have been<br />
criticized by researchers <strong>and</strong> likely overstate the hazard to patients, 6-8 it is undeniable that some<br />
industries, most notably commercial aviation, have safety records far superior to that <strong>of</strong> health<br />
care. One issue we faced in compiling this evidence-based review was the extent to which<br />
specific practices could be identified as playing a direct <strong>and</strong> measurable role in this achievement.<br />
Interestingly, the same issue—ascertaining a causative variable—arose in reviewing the<br />
literature on anesthesia, likely the one field <strong>of</strong> medicine with a safety record that rivals aviation’s<br />
(see also Chapter 56).<br />
25
As outlined in Chapter 24, significant complications attributable to anesthesia have<br />
decreased 9-12 to the point that major morbidity <strong>and</strong> mortality are now too rare to serve as<br />
practical endpoints for measuring the quality <strong>of</strong> anesthesia care, even in large multicenter<br />
studies. 13,14 In attempting to account for this decrease, however, it is very difficult to find<br />
evidence supporting a causative role for even the most plausible c<strong>and</strong>idates, such as widely<br />
utilized intraoperative monitoring st<strong>and</strong>ards. 15 In other words, while the field <strong>of</strong> anesthesia has<br />
clearly made tremendous strides in improving patient safety over the past 50 years, it is hard to<br />
discern a particular, isolated practice that accounts for the clear <strong>and</strong> dramatic secular change in<br />
its safety. While at one level, a pragmatist might argue, “who cares, as long as it’s safe,” trying<br />
to adopt the lessons <strong>of</strong> anesthesia (or for that matter aviation) to the rest <strong>of</strong> health care is made<br />
more challenging by tenuous causality.<br />
Some might argue that, rather than pinpointing specific practices to embrace from other<br />
industries, health care institutions should emulate organizational models that promote safety in<br />
complex, high-risk industries that manage to operate with high reliability. 16 <strong>Analysis</strong> <strong>of</strong> detailed<br />
<strong>and</strong> interesting case studies 17-22 have fueled a school <strong>of</strong> thought known as high reliability theory,<br />
whose proponents suggest a number <strong>of</strong> organizational features that likely reduce the risk <strong>of</strong><br />
“organizational accidents” <strong>and</strong> other hazards. A cogently argued alternative position, <strong>of</strong>ten called<br />
normal accident theory, questions not only these prescriptions for organizational change, but<br />
fundamentally challenges the idea <strong>of</strong> high reliability in certain kinds <strong>of</strong> complex, “tightly<br />
coupled” organizations. 23,24 These competing schools <strong>of</strong> thought <strong>of</strong>fer interesting <strong>and</strong> valuable<br />
insights into the ways that organizational strategies foster safety, while cautioning about the<br />
ever-present threat <strong>of</strong> new sources <strong>of</strong> error that come with increasingly complex human <strong>and</strong><br />
technical organizations. Unfortunately, this rich literature does not permit ready synthesis within<br />
the framework <strong>of</strong> evidence-based medicine, even using the less stringent st<strong>and</strong>ards we adopted in<br />
evaluating non-medical literature (see Chapters 1 <strong>and</strong> 3).<br />
Even the more engineering-oriented <strong>of</strong> the disciplines with potential relevance to patient<br />
safety yielded a surprising lack <strong>of</strong> empirical evaluation <strong>of</strong> safety practices. For instance,<br />
numerous techniques for “human error identification” <strong>and</strong> “error mode prediction” purport to<br />
anticipate important errors <strong>and</strong> develop preventive measures prospectively. 25-27 Their basic<br />
approach consists <strong>of</strong> breaking down the task <strong>of</strong> interest into component processes, <strong>and</strong> then<br />
assigning a measure <strong>of</strong> the likelihood <strong>of</strong> failure to each process. Many <strong>of</strong> the techniques<br />
mentioned in the literature have received little detailed description 25,26 <strong>and</strong> few have received<br />
any formal validation (eg, by comparing predicted failures modes with observed errors). 28,29<br />
Even setting aside dem<strong>and</strong>s for validation, the impact <strong>of</strong> applying these techniques has not been<br />
assessed. Total quality management <strong>and</strong> continuous quality improvement techniques were<br />
championed as important tools for change in health care based on their presumed success in<br />
other industries, but evaluations <strong>of</strong> their impact on health care have revealed little evidence <strong>of</strong><br />
success. 30-33<br />
In the end, we are left with our feet firmly planted in the middle <strong>of</strong> competing paradigms.<br />
One argues that an evidence-based, scientific approach has served health care well <strong>and</strong> should<br />
not be relaxed simply because a popular practice from a “safer” industry sounds attractive. The<br />
other counters that medicine’s slavish devotion to the scientific <strong>and</strong> epidemiologic method has<br />
placed us in a patient safety straightjacket, unable to consider the value <strong>of</strong> practices developed in<br />
other fields because <strong>of</strong> our myopic traditions <strong>and</strong> “reality.”<br />
We see the merits in both arguments. Health care clearly has much to learn from other<br />
industries. Just as physicians must learn the “basic sciences” <strong>of</strong> immunology <strong>and</strong> molecular<br />
biology, providers <strong>and</strong> leaders interested in making health care safer must learn the “basic<br />
26
sciences” <strong>of</strong> organizational theory <strong>and</strong> human factors engineering. Moreover, the “cases”<br />
presented on rounds should, in addition to classical clinical descriptions, also include the tragedy<br />
<strong>of</strong> the Challenger <strong>and</strong> the successes <strong>of</strong> Motorola. On the other h<strong>and</strong>, an unquestioning embrace<br />
<strong>of</strong> dozens <strong>of</strong> promising practices from other fields is likely to be wasteful, distracting, <strong>and</strong><br />
potentially dangerous. We are drawn to a dictum from the Cold War era—“Trust, but verify.”<br />
References<br />
1. Leape LL. Error in medicine. JAMA. 1994;272:1851-1857.<br />
2. Chassin MR. Is health care ready for Six Sigma quality? Milbank Q. 1998;76:565-591,510.<br />
3. Kohn L, Corrigan J, Donaldson M, editors. To Err Is Human: Building a Safer Health<br />
System. Washington, DC: Committee on Quality <strong>of</strong> Health Care in America, Institute <strong>of</strong><br />
Medicine. National Academy Press; 2000.<br />
4. Barach P, Small SD. Reporting <strong>and</strong> preventing medical mishaps: lessons from non-medical<br />
near miss reporting systems. BMJ. 2000;320:759-763.<br />
5. Helmreich RL. On error management: lessons from aviation. BMJ. 2000;320:781-785.<br />
6. Brennan TA. The Institute <strong>of</strong> Medicine report on medical errors–could it do harm? N Engl<br />
J Med. 2000;342:1123-1125.<br />
7. McDonald CJ, Weiner M, Hui SL. Deaths due to medical errors are exaggerated in Institute<br />
<strong>of</strong> Medicine report. JAMA. 2000;284:93-95.<br />
8. Sox HCJ, Woloshin S. How many deaths are due to medical error? Getting the number<br />
right. Eff Clin Pract. 2000;3:277-283.<br />
9. Phillips OC, Capizzi LS. Anesthesia mortality. Clin Anesth. 1974;10:220-244.<br />
10. Deaths during general anesthesia: technology-related, due to human error, or unavoidable?<br />
An ECRI technology assessment. J Health Care Technol. 1985;1:155-175.<br />
11. Keenan RL, Boyan CP. Cardiac arrest due to anesthesia. A study <strong>of</strong> incidence <strong>and</strong> causes.<br />
JAMA. 1985;253:2373-2377.<br />
12. Eichhorn JH. Prevention <strong>of</strong> intraoperative anesthesia accidents <strong>and</strong> related severe injury<br />
through safety monitoring. Anesthesiology. 1989;70:572-577.<br />
13. Cohen MM, Duncan PG, Pope WD, Biehl D, Tweed WA, MacWilliam L, et al. The<br />
Canadian four-centre study <strong>of</strong> anaesthetic outcomes: II. Can outcomes be used to assess the<br />
quality <strong>of</strong> anaesthesia care? Can J Anaesth. 1992;39:430-439.<br />
14. Moller JT, Johannessen NW, Espersen K, Ravlo O, Pedersen BD, Jensen PF, et al.<br />
R<strong>and</strong>omized evaluation <strong>of</strong> pulse oximetry in 20,802 patients: II. Perioperative events <strong>and</strong><br />
postoperative complications. Anesthesiology. 1993;78:445-453.<br />
15. Orkin FK. Practice st<strong>and</strong>ards: the Midas touch or the emperor's new clothes?<br />
Anesthesiology. 1989;70:567-571.<br />
16. Reason J. Human error: models <strong>and</strong> management. BMJ. 2000;320:768-770.<br />
17. Weick KE. Organizational culture as a source <strong>of</strong> high reliability. California management<br />
review. 1987;29:112-127.<br />
18. Roberts KH. Some characteristics <strong>of</strong> high reliability organizations. Berkeley, CA:<br />
Produced <strong>and</strong> distributed by Center for Research in Management, University <strong>of</strong> California,<br />
Berkeley Business School; 1988.<br />
19. LaPorte TR. The United States air traffic control system : increasing reliability in the midst<br />
<strong>of</strong> rapid growth. In: Mayntz R, Hughes TP, editors. The Development <strong>of</strong> Large technical<br />
Systems. Boulder, CO: Westview Press; 1988.<br />
20. Roberts KH. Managing High Reliability Organizations. California Management Review.<br />
1990;32:101-113.<br />
27
21. Roberts KH, Libuser C. From Bhopal to banking: Organizational design can mitigate risk.<br />
Organizational Dynamics. 1993;21:15-26.<br />
22. LaPorte TR, Consolini P. Theoretical <strong>and</strong> operational challenges <strong>of</strong> "high-reliability<br />
organizations": Air-traffic control <strong>and</strong> aircraft carriers. International Journal <strong>of</strong> <strong>Public</strong><br />
Administration. 1998;21:847-852.<br />
23. Perrow C. Normal accidents : Living with High-Risk Technologies. With a New Afterword<br />
<strong>and</strong> a Postscript on the Y2K Problem. Princeton, NJ: Princeton University Press; 1999.<br />
24. Sagan SD. The Limits <strong>of</strong> <strong>Safety</strong>: Organizations, Accidents <strong>and</strong> Nuclear Weapons.<br />
Princeton, NJ: Princeton University Press; 1993.<br />
25. Kirwan B. Human error identification techniques for risk assessment <strong>of</strong> high risk systems–<br />
Part 1: Review <strong>and</strong> evaluation <strong>of</strong> techniques. Appl Ergon. 1998;29:157-177.<br />
26. Kirwan B. Human error identification techniques for risk assessment <strong>of</strong> high risk systems–<br />
Part 2: towards a framework approach. Appl Ergon. 1998;29:299-318.<br />
27. Hollnagel E, Kaarstad M, Lee HC. Error mode prediction. Ergonomics. 1999;42:1457-<br />
1471.<br />
28. Kirwan B, Kennedy R, Taylor-Adams S, Lambert B. The validation <strong>of</strong> three human<br />
reliability quantification techniques–THERP, HEART <strong>and</strong> JHEDI: Part II–Results <strong>of</strong><br />
validation exercise. Appl Ergon. 1997;28:17-25.<br />
29. Stanton NA, Stevenage SV. Learning to predict human error: issues <strong>of</strong> acceptability,<br />
reliability <strong>and</strong> validity. Ergonomics. 1998;41:1737-1756.<br />
30. Blumenthal D, Kilo CM. A report card on continuous quality improvement. Milbank Q.<br />
1998;76:625-48,511.<br />
31. Gerowitz MB. Do TQM interventions change management culture? Findings <strong>and</strong><br />
implications. Qual Manag Health Care. 1998;6:1-11.<br />
32. Shortell SM, Bennett CL, Byck GR. Assessing the impact <strong>of</strong> continuous quality<br />
improvement on clinical practice: what it will take to accelerate progress. Milbank Q.<br />
1998;76:593-624,510.<br />
33. Shortell SM, Jones RH, Rademaker AW, Gillies RR, Dranove DS, Hughes EF, et al.<br />
Assessing the impact <strong>of</strong> total quality management <strong>and</strong> organizational culture on multiple<br />
outcomes <strong>of</strong> care for coronary artery bypass graft surgery patients. Med Care.<br />
2000;38:207-217.<br />
28
Chapter 3. Evidence-based Review Methodology<br />
Definition <strong>and</strong> Scope<br />
For this project the UCSF-Stanford Evidence-based Practice Center (EPC) defined a<br />
patient safety practice as “a type <strong>of</strong> process or structure whose application reduces the<br />
probability <strong>of</strong> adverse events resulting from exposure to the health care system across a range <strong>of</strong><br />
conditions or procedures.” Examples <strong>of</strong> practices that meet this definition include computerized<br />
physician order entry (Chapter 6), thromboembolism prophylaxis in hospitalized patients<br />
(Chapter 31), strategies to reduce falls among hospitalized elders (Chapter 26), <strong>and</strong> novel<br />
education strategies such as the application <strong>of</strong> “crew resource management” to train operating<br />
room staff (Chapter 44). By contrast, practices that are disease-specific <strong>and</strong>/or directed at the<br />
underlying disease <strong>and</strong> its complications (eg, use <strong>of</strong> aspirin or beta-blockers to treat acute<br />
myocardial infarction) rather than complications <strong>of</strong> medical care are not included as “patient<br />
safety practices.” Further discussion <strong>of</strong> these issues can be found in Chapter 1.<br />
In some cases, the distinction between patient safety <strong>and</strong> more general quality<br />
improvement strategies is difficult to discern. Quality improvement practices may also qualify as<br />
patient safety practices when the current level <strong>of</strong> quality is considered “unsafe,” but st<strong>and</strong>ards to<br />
measure safety are <strong>of</strong>ten variable, difficult to quantify, <strong>and</strong> change over time. For example, what<br />
constitutes an “adequate” or “safe” level <strong>of</strong> accuracy in electrocardiogram or radiograph<br />
interpretation? <strong>Practices</strong> to improve performance to at least the “adequate” threshold may<br />
reasonably be considered safety practices because they decrease the number <strong>of</strong> diagnostic errors<br />
<strong>of</strong> omission. On the other h<strong>and</strong>, we considered practices whose main intent is to improve<br />
performance above this threshold to be quality improvement practices. An example <strong>of</strong> the latter<br />
might be the use <strong>of</strong> computer algorithms to improve the sensitivity <strong>of</strong> screening mammography. 1<br />
We generally included practices that involved acute hospital care or care at the interface<br />
between inpatient <strong>and</strong> outpatient settings. This focus reflects the fact that the majority <strong>of</strong> the<br />
safety literature relates to acute care <strong>and</strong> the belief that systems changes may be more effectively<br />
achieved in the more controlled environment <strong>of</strong> the hospital. However, practices that might be<br />
applicable in settings in addition to the hospital were not excluded from consideration. For<br />
example, the Report includes practices for preventing decubitus ulcers (Chapter 27) that could be<br />
applied in nursing homes as well as hospitals.<br />
The EPC team received input regarding the scope <strong>of</strong> the Report from the Agency for<br />
Healthcare Research <strong>and</strong> Quality (AHRQ), which commissioned the report. In addition, the EPC<br />
team participated in the public meeting <strong>of</strong> the National Quality Forum (NQF) Safe <strong>Practices</strong><br />
Committee on January 26, 2001. The NQF was formed in 1999 by consumer, purchaser,<br />
provider, health plan, <strong>and</strong> health service research organizations to create a national strategy for<br />
quality improvement. Members <strong>of</strong> the Safe <strong>Practices</strong> Committee collaborated with the EPC team<br />
to develop the scope <strong>of</strong> work that would eventually become this Report.<br />
Organization by Domains<br />
To facilitate identification <strong>and</strong> evaluation <strong>of</strong> potential patient safety practices, we divided<br />
the content for the project into different domains. Some cover “content areas” (eg, adverse drug<br />
events, nosocomial infections, <strong>and</strong> complications <strong>of</strong> surgery). Other domains involve<br />
identification <strong>of</strong> practices within broad (primarily “non-medical”) disciplines likely to contain<br />
promising approaches to improving patient safety (eg, information technology, human factors<br />
research, organizational theory). The domains were derived from a general reading <strong>of</strong> the<br />
29
literature <strong>and</strong> were meant to be as inclusive as possible. The list underwent review for<br />
completeness by patient safety experts, clinician-researchers, AHRQ, <strong>and</strong> the NQF Safe<br />
<strong>Practices</strong> Committee. For each domain we selected a team <strong>of</strong> author/collaborators with expertise<br />
in the relevant subject matter <strong>and</strong>/or familiarity with the techniques <strong>of</strong> evidence-based review<br />
<strong>and</strong> technology appraisal. The authors, all <strong>of</strong> whom are affiliated with major academic centers<br />
around the United States, are listed on page 9-13.<br />
Identification <strong>of</strong> <strong>Safety</strong> <strong>Practices</strong> for Evaluation<br />
Search Strategy<br />
The UCSF-Stanford Evidence-based Practice Center Coordinating Team (“The Editors”)<br />
provided general instructions (Table 3.1) to the teams <strong>of</strong> authors regarding search strategies for<br />
identifying safety practices for evaluation. As necessary, the Editors provided additional<br />
guidance <strong>and</strong> supplementary searches <strong>of</strong> the literature.<br />
Table 3.1. Search strategy recommended by coordinating team to<br />
participating reviewers<br />
a) Electronic bibliographic databases. All searches must include systematic<br />
searches <strong>of</strong> MEDLINE <strong>and</strong> the Cochrane Library. For many topics it will<br />
be necessary to include other databases such as the Cumulative Index to<br />
Nursing & Allied Health (CINAHL), PsycLit (PsycINFO), the Institute for<br />
Scientific Information’s Science Citation Index Exp<strong>and</strong>ed, Social Sciences<br />
Citation Index, Arts & Humanities Citation Index, INSPEC (physics,<br />
electronics <strong>and</strong> computing), <strong>and</strong> ABI/INFORM (business, management,<br />
finance, <strong>and</strong> economics).<br />
b) H<strong>and</strong>-searches <strong>of</strong> bibliographies <strong>of</strong> retrieved articles <strong>and</strong> tables <strong>of</strong><br />
contents <strong>of</strong> key journals.<br />
c) Grey literature. For many topics it will necessary to review the “grey<br />
literature,” such as conference proceedings, institutional reports, doctoral<br />
theses, <strong>and</strong> manufacturers’ reports.<br />
d) Consultation with experts or workers in the field.<br />
To meet the early dissemination date m<strong>and</strong>ated by AHRQ, the Editors did not require<br />
authors to search for non-English language articles or to use EMBASE. These were not<br />
specifically excluded, however, <strong>and</strong> authoring teams could include non-English language articles<br />
that addressed important aspects <strong>of</strong> a topic if they had translation services at their disposal. The<br />
Editors did not make recommendations on limiting database searches based on publication date.<br />
For this project it was particularly important to identify systematic reviews related to patient<br />
safety topics. Published strategies for retrieving systematic reviews have used proprietary<br />
MEDLINE interfaces (eg, OVID, SilverPlatter) that are not uniformly available. Moreover, the<br />
performance characteristics <strong>of</strong> these search strategies is unknown. 2,3 Therefore, these strategies<br />
were not explicitly recommended. The Editors provided authors with a search algorithm<br />
(available upon request) that uses PubMed, the freely available search interface from the<br />
National Library <strong>of</strong> Medicine, designed to retrieve systematic reviews with high sensitivity<br />
without overwhelming users with “false positive” hits. 4<br />
30
The Editors also performed independent searches <strong>of</strong> bibliographic databases <strong>and</strong> grey<br />
literature for selected topics. 5 Concurrently, the EPC collaborated with NQF to solicit<br />
information about evidence-based practices from NQF members, <strong>and</strong> consulted with the<br />
project’s Advisory Panel (page 31), whose members provided additional literature to review.<br />
Inclusion <strong>and</strong> Exclusion Criteria<br />
The EPC established criteria for selecting which <strong>of</strong> the identified safety practices<br />
warranted evaluation. The criteria address the applicability <strong>of</strong> the practice across a range <strong>of</strong><br />
conditions or procedures <strong>and</strong> the available evidence <strong>of</strong> the practices’ efficacy or effectiveness.<br />
Table 3.2. Inclusion/Exclusion criteria for practices<br />
Inclusion Criteria<br />
1. The practice can be applied in the hospital setting or at the interface<br />
between inpatient <strong>and</strong> outpatient settings AND can be applied to a broad<br />
range <strong>of</strong> health care conditions or procedures.<br />
2. Evidence for the safety practice includes at least one study with a Level 3<br />
or higher study design AND a Level 2 outcome measure. For practices not<br />
specifically related to diagnostic or therapeutic interventions, a Level 3<br />
outcome measure is adequate. (See Table 3.3 for definition <strong>of</strong> “Levels”).<br />
Exclusion Criterion<br />
1. No study <strong>of</strong> the practice meets the methodologic criteria above.<br />
<strong>Practices</strong> that have only been studied outside the hospital setting or in patients with<br />
specific conditions or undergoing specific procedures were included if the authors <strong>and</strong> Editors<br />
agreed that the practices could reasonably be applied in the hospital setting <strong>and</strong> across a range <strong>of</strong><br />
conditions or procedures. To increase the number <strong>of</strong> potentially promising safety practices<br />
adapted from outside the field <strong>of</strong> medicine, we included evidence from studies that used less<br />
rigorous measures <strong>of</strong> patient safety as long as the practices did not specifically relate to<br />
diagnostic or therapeutic interventions. These criteria facilitated the inclusion <strong>of</strong> areas such as<br />
teamwork training (Chapter 44) <strong>and</strong> methods to improve information transfer (Chapter 42).<br />
Each practice’s level <strong>of</strong> evidence for efficacy or effectiveness was assessed in terms <strong>of</strong><br />
study design (Table 3.3) <strong>and</strong> study outcomes (Table 3.4). The Editors created the following<br />
hierarchies by modifying existing frameworks for evaluating evidence 6-8 <strong>and</strong> incorporating<br />
recommendations from numerous other sources relevant to evidence synthesis. 9-25<br />
31
Table 3.3. Hierarchy <strong>of</strong> study designs*<br />
Level 1. R<strong>and</strong>omized controlled trials – includes quasi-r<strong>and</strong>omized processes<br />
such as alternate allocation<br />
Level 2. Non-r<strong>and</strong>omized controlled trial – a prospective (pre-planned) study,<br />
with predetermined eligibility criteria <strong>and</strong> outcome measures.<br />
Level 3. Observational studies with controls – includes retrospective,<br />
interrupted time series (a change in trend attributable to the<br />
intervention), case-control studies, cohort studies with controls, <strong>and</strong><br />
health services research that includes adjustment for likely<br />
confounding variables<br />
Level 4. Observational studies without controls (eg, cohort studies without<br />
controls <strong>and</strong> case series)<br />
* Systematic reviews <strong>and</strong> meta-analyses were assigned to the highest level<br />
study design included in the review, followed by an “A” (eg, a systematic<br />
review that included at least one r<strong>and</strong>omized controlled trial was designated<br />
“Level 1A”)<br />
Table 3.4. Hierarchy <strong>of</strong> outcome measures<br />
Level 1. Clinical outcomes - morbidity, mortality, adverse events<br />
Level 2. Surrogate outcomes - observed errors, intermediate outcomes (eg,<br />
laboratory results) with well-established connections to the clinical<br />
outcomes <strong>of</strong> interest (usually adverse events).<br />
Level 3. Other measurable variables with an indirect or unestablished<br />
connection to the target safety outcome (eg, pre-test/post-test after<br />
an educational intervention, operator self-reports in different<br />
experimental situations)<br />
Level 4. No outcomes relevant to decreasing medical errors <strong>and</strong>/or adverse<br />
events (eg, study with patient satisfaction as only measured<br />
outcome; article describes an approach to detecting errors but reports<br />
no measured outcomes)<br />
Implicit in this hierarchy <strong>of</strong> outcome measures is that surrogate or intermediate outcomes<br />
(Level 2) have an established relationship to the clinical outcomes (Level 1) <strong>of</strong> interest. 26<br />
Outcomes that are relevant to patient safety but have not been associated with morbidity or<br />
mortality were classified as Level 3.<br />
Exceptions to EPC Criteria<br />
Some safety practices did not meet the EPC inclusion criteria because <strong>of</strong> the paucity <strong>of</strong><br />
evidence regarding efficacy or effectiveness, but were included in the Report because <strong>of</strong> their<br />
face validity (ie, an informed reader might reasonably expect them to be evaluated; see also<br />
Chapter 1). The reviews <strong>of</strong> these practices clearly identify the quality <strong>of</strong> evidence culled from<br />
medical <strong>and</strong> non-medical fields.<br />
32
Evaluation <strong>of</strong> <strong>Safety</strong> <strong>Practices</strong><br />
For each practice, authors were instructed to research the literature for information on:<br />
• prevalence <strong>of</strong> the problem targeted by the practice<br />
• severity <strong>of</strong> the problem targeted by the practice<br />
• the current utilization <strong>of</strong> the practice<br />
• evidence on efficacy <strong>and</strong>/or effectiveness <strong>of</strong> the practice<br />
• the practice’s potential for harm<br />
• data on cost if available<br />
• implementation issues<br />
These elements were incorporated into a template in an effort to create as much<br />
uniformity across chapters as possible, especially given the widely disparate subject matter <strong>and</strong><br />
quality <strong>of</strong> evidence. Since the amount <strong>of</strong> material for each practice was expected to, <strong>and</strong> did,<br />
vary substantially, the Editors provided general guidance on what was expected for each<br />
element, with particular detail devoted to the protocol for searching <strong>and</strong> reporting evidence<br />
related to efficacy <strong>and</strong>/or effectiveness <strong>of</strong> the practice.<br />
The protocol outlined the search, the threshold for study inclusion, the elements to<br />
abstract from studies, <strong>and</strong> guidance on reporting information from each study. Authors were<br />
asked to review articles from their search to identify practices, <strong>and</strong> retain those with the better<br />
study designs. More focused searches were performed depending on the topic. The threshold for<br />
study inclusion related directly to study design. Authors were asked to use their judgment in<br />
deciding whether the evidence was sufficient at a given level <strong>of</strong> study design or whether the<br />
evidence from the next level needed to be reviewed. At a minimum, the Editors suggested that<br />
there be at least 2 studies <strong>of</strong> adequate quality to justify excluding discussion <strong>of</strong> studies <strong>of</strong> lower<br />
level designs. Thus inclusion <strong>of</strong> 2 adequate clinical trials (Level 1 design) were necessary in<br />
order to exclude available evidence from prospective, non-r<strong>and</strong>omized trials (Level 2) on the<br />
same topic.<br />
The Editors provided instructions for abstracting each article that met the inclusion<br />
criteria based on study design. For each study, a required set <strong>of</strong> 10 abstraction elements (Table<br />
3.5) was specified. Authors received a detailed explanation <strong>of</strong> each required abstraction element,<br />
as well as complete abstraction examples for the 3 types <strong>of</strong> study design (Levels 1A, 1, <strong>and</strong> 3;<br />
Level 2 was not included since the information collected was same as Level 1). Research teams<br />
were encouraged to abstract any additional elements relevant to the specific subject area.<br />
33
Table 3.5. Ten Required Abstraction Elements<br />
1. Bibliographic information according to AMA Manual <strong>of</strong> Style: title, authors, date<br />
<strong>of</strong> publication, source<br />
2. Level <strong>of</strong> study design (eg, Level 1-3 for studies providing information for<br />
effectiveness; Level 4 if needed for relevant additional information) with<br />
descriptive material as follows:<br />
Descriptors<br />
For Level 1 Systematic Reviews Only<br />
(a) Identifiable description <strong>of</strong> methods indicating sources <strong>and</strong> methods <strong>of</strong><br />
searching for articles.<br />
(b) Stated inclusion <strong>and</strong> exclusion criteria for articles: yes/no.<br />
(c) Scope <strong>of</strong> literature included in study.<br />
For Level 1 or 2 Study Designs (Not Systematic Reviews)<br />
(a) Blinding: blinded, blinded (unclear), or unblinded<br />
(b) Describe comparability <strong>of</strong> groups at baseline – ie, was distribution <strong>of</strong> potential<br />
confounders at baseline equal? If no, which confounders were not equal?<br />
(c) Loss to follow-up overall: percent <strong>of</strong> total study population lost to follow-up.<br />
For Level 3 Study Design (Not Systematic Reviews)<br />
(a) Description <strong>of</strong> study design (eg, case-control, interrupted time series).<br />
(b) Describe comparability <strong>of</strong> groups at baseline – ie, was distribution <strong>of</strong> potential<br />
confounders at baseline equal? If no, which confounders were not equal?<br />
(c) <strong>Analysis</strong> includes adjustment for potential confounders: yes/no. If yes,<br />
adjusted for what confounders?<br />
3. Description <strong>of</strong> intervention (as specific as possible)<br />
4. Description <strong>of</strong> study population(s) <strong>and</strong> setting(s)<br />
5. Level <strong>of</strong> relevant outcome measure(s) (eg, Levels 1-4)<br />
6. Description <strong>of</strong> relevant outcome measure(s)<br />
7. Main Results: effect size with confidence intervals<br />
8. Information on unintended adverse (or beneficial) effects <strong>of</strong> practice<br />
9. Information on cost <strong>of</strong> practice<br />
10. Information on implementation <strong>of</strong> practice (information that might be <strong>of</strong> use in<br />
whether to <strong>and</strong>/or how to implement the practice – eg, known barriers to<br />
implementation)<br />
We present the salient elements <strong>of</strong> each included study (eg, study design,<br />
population/setting, intervention details, results) in text or tabular form. In addition, we asked<br />
authors to highlight weaknesses <strong>and</strong> biases <strong>of</strong> studies where the interpretation <strong>of</strong> the results<br />
might be substantially affected. Authors were not asked to formally synthesize or combine (eg,<br />
perform a meta-analysis) the evidence across studies for the Report.<br />
34
Review Process<br />
Authors submitted work to the Editors in 2 phases. In the first phase (“Identification <strong>of</strong><br />
<strong>Safety</strong> <strong>Practices</strong> for Evaluation”), which was submitted approximately 6 weeks after authors<br />
were commissioned, authoring teams provided their search strategies, citations, <strong>and</strong> a<br />
preliminary list <strong>of</strong> patient safety practices to be reviewed. In the subsequent phase (“Evaluation<br />
<strong>of</strong> <strong>Safety</strong> <strong>Practices</strong>”), due approximately 12 weeks after commissioning, authors first submitted a<br />
draft chapter for each topic, completed abstraction forms, <strong>and</strong>—after iterative reviews <strong>and</strong><br />
revisions—a final chapter.<br />
Identification <strong>of</strong> <strong>Safety</strong> <strong>Practices</strong> for Evaluation<br />
The Editors <strong>and</strong> the Advisory Panel reviewed the list <strong>of</strong> domains <strong>and</strong> practices to identify<br />
gaps in coverage. In addition, the Editors reviewed final author-submitted lists <strong>of</strong> excluded<br />
practices along with justifications for exclusion (eg, insufficient research design, insufficient<br />
outcomes, practice is unique to a single disease process). When there were differences in opinion<br />
as to whether a practice actually met the inclusion or exclusion criteria, the Editors made a final<br />
disposition after consulting with the author(s). The final practice list, in the form <strong>of</strong> a Table <strong>of</strong><br />
Contents for the Report, was circulated to AHRQ <strong>and</strong> the NQF Safe <strong>Practices</strong> Committee for<br />
comment.<br />
Evaluation <strong>of</strong> <strong>Safety</strong> <strong>Practices</strong><br />
Chapters were reviewed by the editorial team (The EPC Coordinating Team Editors <strong>and</strong><br />
our Managing Editor) <strong>and</strong> queries were relayed to the authors, <strong>of</strong>ten requesting further<br />
refinement <strong>of</strong> the analysis or expansion <strong>of</strong> the results <strong>and</strong> conclusions. After all chapters were<br />
completed, the entire Report was edited to eliminate redundant material <strong>and</strong> ensure that the focus<br />
remained on the evidence regarding safety practices. Near the end <strong>of</strong> the review process,<br />
chapters were distributed to the Advisory Panel for comments, many <strong>of</strong> which were<br />
incorporated. Once the content was finalized, the Editors analyzed <strong>and</strong> ranked the practices using<br />
a methodology described in Chapter 56. The results <strong>of</strong> these summaries <strong>and</strong> rankings are<br />
presented in Part V <strong>of</strong> the Report.<br />
35
References<br />
1. Thurfjell E, Thurfjell MG, Egge E, Bjurstam N. Sensitivity <strong>and</strong> specificity <strong>of</strong> computerassisted<br />
breast cancer detection in mammography screening. Acta Radiol. 1998;39:384-<br />
388.<br />
2. Hunt DL, McKibbon KA. Locating <strong>and</strong> appraising systematic reviews Ann Intern Med.<br />
1997;126:532-538.<br />
3. Glanville J, Lefebvre C. Identifying systematic reviews: key resources. ACP J Club.<br />
2000;132:A11-A12.<br />
4. Shojania KG, Bero L. Taking advantage <strong>of</strong> the explosion <strong>of</strong> systematic reviews: an<br />
efficient MEDLINE search strategy. Eff Clin Pract. 2001;4: in press.<br />
5. McAuley L, Pham B, Tugwell P, Moher D. Does the inclusion <strong>of</strong> grey literature influence<br />
estimates <strong>of</strong> intervention effectiveness reported in meta-analyses? Lancet. 2000;356:1228-<br />
1231.<br />
6. Harris RP, Helf<strong>and</strong> M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, et al. Current<br />
methods <strong>of</strong> the U.S. Preventive Services Task Force. A review <strong>of</strong> the process. Am J Prev<br />
Med. 2001;20:21-35.<br />
7. Guyatt G, Schunemann H, Cook D, Jaeschke R, Pauker S, Bucher H. Grades <strong>of</strong><br />
recommendation for antithrombotic agents. Chest.2001;119:3S-7S.<br />
8. Muir Gray JA, Haynes RB, Sackett DL, Cook DJ, Guyatt GH. Transferring evidence from<br />
research into practice: 3. Developing evidence-based clinical policy. ACP J Club.<br />
1997;126:A14-A16.<br />
9. Guyatt GH, Tugwell PX, Feeny DH, Drummond MF, Haynes RB. The role <strong>of</strong> before-after<br />
studies <strong>of</strong> therapeutic impact in the evaluation <strong>of</strong> diagnostic technologies. J Chronic Dis.<br />
1986;39:295-304.<br />
10. Davis CE. Generalizing from clinical trials. Control Clin Trials. 1994;15:11-4.<br />
11. Bailey KR. Generalizing the results <strong>of</strong> r<strong>and</strong>omized clinical trials. Control Clin Trials.<br />
1994;15:15-23.<br />
12. Oxman AD, Cook DJ, Guyatt GH. Users' guides to the medical literature. VI. How to use<br />
an overview. Evidence-Based Medicine Working Group. JAMA. 1994;272:1367-1371.<br />
13. Levine M, Walter S, Lee H, Haines T, Holbrook A, Moyer V. Users' guides to the medical<br />
literature. IV. How to use an article about harm. Evidence-Based Medicine Working<br />
Group. JAMA. 1994;271:1615-1619.<br />
14. Guyatt GH, Sackett DL, Sinclair JC, Hayward R, Cook DJ, Cook RJ. Users' guides to the<br />
medical literature. IX. A method for grading health care recommendations. Evidence-<br />
Based Medicine Working Group. JAMA. 1995;274:1800-1804.<br />
15. Cook DJ, Guyatt GH, Laupacis A, Sackett DL, Goldberg RJ. Clinical recommendations<br />
using levels <strong>of</strong> evidence for antithrombotic agents. Chest. 1995;108:227S-230S.<br />
16. Naylor CD, Guyatt GH. Users' guides to the medical literature. X. How to use an article<br />
reporting variations in the outcomes <strong>of</strong> health services. The Evidence-Based Medicine<br />
Working Group. JAMA. 1996;275:554-558.<br />
17. Moher D, Jadad AR, Tugwell P. Assessing the quality <strong>of</strong> r<strong>and</strong>omized controlled trials.<br />
Current issues <strong>and</strong> future directions. Int J Technol Assess Health Care. 1996;12:195-208.<br />
18. Mulrow C, Langhorne P, Grimshaw J. Integrating heterogeneous pieces <strong>of</strong> evidence in<br />
systematic reviews. Ann Intern Med. 1997;127:989-995.<br />
19. Jadad AR, Cook DJ, Browman GP. A guide to interpreting discordant systematic reviews.<br />
CMAJ. 1997;156:1411-1416.<br />
36
20. Juni P, Witschi A, Bloch R, Egger M. The hazards <strong>of</strong> scoring the quality <strong>of</strong> clinical trials<br />
for meta-analysis. JAMA. 1999;282:1054-1060.<br />
21. McKee M, Britton A, Black N, McPherson K, S<strong>and</strong>erson C, Bain C. Methods in health<br />
services research. Interpreting the evidence: choosing between r<strong>and</strong>omised <strong>and</strong> nonr<strong>and</strong>omised<br />
studies. BMJ. 1999;319:312-315.<br />
22. Bucher HC, Guyatt GH, Cook DJ, Holbrook A, McAlister FA. Users' guides to the medical<br />
literature: XIX. Applying clinical trial results. A. How to use an article measuring the<br />
effect <strong>of</strong> an intervention on surrogate end points. Evidence-Based Medicine Working<br />
Group. JAMA. 1999;282:771-778.<br />
23. Concato J, Shah N, Horwitz RI. R<strong>and</strong>omized, controlled trials, observational studies, <strong>and</strong><br />
the hierarchy <strong>of</strong> research designs. N Engl J Med. 2000;342:1887-1892.<br />
24. Benson K, Hartz AJ. A comparison <strong>of</strong> observational studies <strong>and</strong> r<strong>and</strong>omized, controlled<br />
trials. N Engl J Med. 2000;342:1878-1886.<br />
25. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis<br />
<strong>of</strong> observational studies in epidemiology: a proposal for reporting. Meta-analysis Of<br />
Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000;283:2008-2012.<br />
26. Gotzsche PC, Liberati A, Torri V, Rossetti L. Beware <strong>of</strong> surrogate outcome measures. Int J<br />
Technol Assess Health Care. 1996;12:238-246.<br />
37
PART II. REPORTING AND RESPONDING TO PATIENT SAFETY PROBLEMS<br />
Chapter 4. Incident Reporting<br />
Chapter 5. Root Cause <strong>Analysis</strong><br />
39
Chapter 4. Incident Reporting<br />
Heidi Wald, MD<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Errors in medical care are discovered through a variety <strong>of</strong> mechanisms. Historically,<br />
medical errors were revealed retrospectively through morbidity <strong>and</strong> mortality committees <strong>and</strong><br />
malpractice claims data. Prominent studies <strong>of</strong> medical error have used retrospective chart review<br />
to quantify adverse event rates. 1,2 While collection <strong>of</strong> data in this manner yields important<br />
epidemiologic information, it is costly <strong>and</strong> provides little insight into potential error reduction<br />
strategies. Moreover, chart review only detects documented adverse events <strong>and</strong> <strong>of</strong>ten does not<br />
capture information regarding their causes. Important errors that produce no injury may go<br />
completely undetected by this method. 3-6<br />
Computerized surveillance may also play a role in uncovering certain types <strong>of</strong> errors. For<br />
instance, medication errors may be discovered through a search for naloxone orders for<br />
hospitalized patients, as they presumably reflect the need to reverse overdose <strong>of</strong> prescribed<br />
narcotics. 7,8 Several studies have demonstrated success with computerized identification <strong>of</strong><br />
adverse drug events. 9-11<br />
Complex, high-risk industries outside <strong>of</strong> health care, including aviation, nuclear power,<br />
petrochemical processing, steel production, <strong>and</strong> military operations, have successfully developed<br />
incident reporting systems for serious accidents <strong>and</strong> important “near misses.” 6 Incident reporting<br />
systems cannot provide accurate epidemiologic data, as the reported incidents likely<br />
underestimate the numerator, <strong>and</strong> the denominator (all opportunities for incidents) remains<br />
unknown.<br />
Given the limited availability <strong>of</strong> sophisticated clinical computer systems <strong>and</strong> the<br />
tremendous resources required to conduct comprehensive chart reviews, incident reporting<br />
systems remain an important <strong>and</strong> relatively inexpensive means <strong>of</strong> capturing data on errors <strong>and</strong><br />
adverse events in medicine. Few rigorous studies have analyzed the benefits <strong>of</strong> incident<br />
reporting. This chapter reviews only the literature evaluating the various systems <strong>and</strong> techniques<br />
for collecting error data in this manner, rather than the benefit <strong>of</strong> the practice itself. This decision<br />
reflects our acknowledgment that incident reporting has clearly played a beneficial role in other<br />
high-risk industries. 6 The decision also stems from our recognition that a measurable impact <strong>of</strong><br />
incident reporting on clinical outcomes is unlikely because there is no st<strong>and</strong>ard practice by which<br />
institutions h<strong>and</strong>le these reports.<br />
Practice Description<br />
Flanagan first described the critical incident technique in 1954 to examine military<br />
aircraft training accidents. 12 <strong>Critical</strong> incident reporting involves the identification <strong>of</strong> preventable<br />
incidents (ie, occurrences that could have led, or did lead, to an undesirable outcome 13 ) reported<br />
by personnel directly involved in the process in question at the time <strong>of</strong> discovery <strong>of</strong> the event.<br />
The goal <strong>of</strong> critical incident monitoring is not to gather epidemiologic data per se, but rather to<br />
gather qualitative data. Nonetheless, if a pattern <strong>of</strong> errors seems to emerge, prospective studies<br />
can be undertaken to test epidemiologic hypotheses. 14<br />
41
Incident reports may target events in any or all <strong>of</strong> 3 basic categories: adverse events, “no<br />
harm events,” <strong>and</strong> “near misses.” For example, anaphylaxis to penicillin clearly represents an<br />
adverse event. Intercepting the medication order prior to administration would constitute a near<br />
miss. By contrast, if a patient with a documented history <strong>of</strong> anaphylaxis to penicillin received a<br />
penicillin-like antibiotic (eg, a cephalosporin) but happened not to experience an allergic<br />
reaction, it would constitute a no harm event, not a near miss. In other words, when an error does<br />
not result in an adverse event for a patient, because the error was “caught,” it is a near miss; if<br />
the absence <strong>of</strong> injury is owed to chance it is a no harm event. Broadening the targets <strong>of</strong> incident<br />
reporting to include no harm events <strong>and</strong> near misses <strong>of</strong>fers several advantages. These events<br />
occur 3 to 300 times more <strong>of</strong>ten than adverse events, 5,6 they are less likely to provoke guilt or<br />
other psychological barriers to reporting, 6 <strong>and</strong> they involve little medico-legal risk. 14 In addition,<br />
hindsight bias 15 is less likely to affect investigations <strong>of</strong> no harm events <strong>and</strong> near misses. 6,14<br />
Barach <strong>and</strong> Small describe the characteristics <strong>of</strong> incident reporting systems in nonmedical<br />
industries. 6 Established systems share the following characteristics:<br />
• they focus on near misses<br />
• they provide incentives for voluntary reporting;<br />
• they ensure confidentiality; <strong>and</strong><br />
• they emphasize systems approaches to error analysis.<br />
The majority <strong>of</strong> these systems were m<strong>and</strong>ated by Federal regulation, <strong>and</strong> provide for<br />
voluntary reporting. All <strong>of</strong> the systems encourage narrative description <strong>of</strong> the event. Reporting is<br />
promoted by providing incentives including:<br />
• immunity;<br />
• confidentiality;<br />
• outsourcing <strong>of</strong> report collation;<br />
• rapid feedback to all involved <strong>and</strong> interested parties; <strong>and</strong><br />
• sustained leadership support. 6<br />
Incident reporting in medicine takes many forms. Since 1975, the US Food <strong>and</strong> Drug<br />
Administration (FDA) has m<strong>and</strong>ated reporting <strong>of</strong> major blood transfusion reactions, focusing on<br />
preventable deaths <strong>and</strong> serious injuries. 16 Although the critical incident technique found some<br />
early applications in medicine, 17,18 its current use is largely attributable to Cooper’s introduction<br />
<strong>of</strong> incident reporting to anesthesia in 1978, 19 conducting retrospective interviews with<br />
anesthesiologists about preventable incidents or errors that occurred while patients were under<br />
their care. Recently, near miss <strong>and</strong> adverse event reporting systems have proliferated in single<br />
institution settings (such as in intensive care units (ICUs) 20,21 ), regional settings (such as the New<br />
York State transfusion system 22 ), <strong>and</strong> for national surveillance (eg, the National Nosocomial<br />
Infections Surveillance System administered by the Federal Centers for Disease Control <strong>and</strong><br />
Prevention.) 23<br />
All <strong>of</strong> the above examples focus on types <strong>of</strong> events (transfusion events or nosocomial<br />
infections) or areas <strong>of</strong> practice (ICUs). Incident reporting in hospitals cuts a wider swath,<br />
capturing errors <strong>and</strong> departures from expected procedures or outcomes (Table 4.1). However,<br />
because risk management departments tend to oversee incident reporting systems in some<br />
capacity, these systems more <strong>of</strong>ten focus on incident outcomes, not categories. Few data describe<br />
the operation <strong>of</strong> these institution-specific systems, but underreporting appears endemic. 24<br />
42
In 1995, hospital-based surveillance was m<strong>and</strong>ated by the Joint Commission on the<br />
Accreditation <strong>of</strong> Healthcare Organizations (JCAHO) 26 because <strong>of</strong> a perception that incidents<br />
resulting in harm were occurring frequently. 28 JCAHO employs the term sentinel event in lieu <strong>of</strong><br />
critical incident, <strong>and</strong> defines it as follows:<br />
An unexpected occurrence involving death or serious physical or psychological<br />
injury, or the risk there<strong>of</strong>. Serious injury specifically includes loss <strong>of</strong> limb or<br />
function. The phrase “or the risk there<strong>of</strong>” includes any process variation for<br />
which a recurrence would carry a significant chance <strong>of</strong> a serious adverse<br />
outcome. 26<br />
As one component <strong>of</strong> its Sentinel Event Policy, JCAHO created a Sentinel Event<br />
Database. The JCAHO database accepts voluntary reports <strong>of</strong> sentinel events from member<br />
institutions, patients <strong>and</strong> families, <strong>and</strong> the press. 26 The particulars <strong>of</strong> the reporting process are left<br />
to the member health care organizations. JCAHO also m<strong>and</strong>ates that accredited hospitals<br />
perform root cause analysis (see Chapter 5) <strong>of</strong> important sentinel events. Data on sentinel events<br />
are collated, analyzed, <strong>and</strong> shared through a website, 29 an online publication, 30 <strong>and</strong> its newsletter<br />
Sentinel Event Perspectives. 31<br />
Another example <strong>of</strong> a national incident reporting system is the Australian Incident<br />
Monitoring Study (AIMS), under the auspices <strong>of</strong> the Australian <strong>Patient</strong> <strong>Safety</strong> Foundation. 13<br />
Investigators created an anonymous <strong>and</strong> voluntary near miss <strong>and</strong> adverse event reporting system<br />
for anesthetists in Australia. Ninety participating hospitals <strong>and</strong> practices named on-site<br />
coordinators. The AIMS group developed a form that was distributed to participants. The form<br />
contained instructions, definitions, space for narrative <strong>of</strong> the event, <strong>and</strong> structured sections to<br />
record the anesthesia <strong>and</strong> procedure, demographics about the patient <strong>and</strong> anesthetist, <strong>and</strong> what,<br />
when, why, where, <strong>and</strong> how the event occurred. The results <strong>of</strong> the first 2000 reports were<br />
published together, following a special symposium. 32<br />
The experiences <strong>of</strong> the JCAHO Sentinel Event Database <strong>and</strong> the Australian Incident<br />
Monitoring Study are explored further below.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The true prevalence <strong>of</strong> events appropriate for incident reporting is impossible to estimate with<br />
any accuracy, as it includes actual adverse events as well as near misses <strong>and</strong> no harm events. The<br />
Aviation <strong>Safety</strong> Reporting System (ASRS), a national reporting system for near misses in the<br />
airline industry,33,34 currently processes approximately 30,000 reports annually,35 exceeding<br />
by many orders <strong>of</strong> magnitude the total number <strong>of</strong> airline accidents each year.34 The number <strong>of</strong><br />
reports submitted to a comparable system in health care would presumably number in the<br />
millions if all adverse events, no harm events, <strong>and</strong> near misses were captured.<br />
By contrast, over 6 years <strong>of</strong> operation, the JCAHO Sentinel Event Database has captured<br />
only 1152 events, 62% <strong>of</strong> which occurred in general hospitals. Two-thirds <strong>of</strong> the events were<br />
self-reported by institutions, with the remainder coming from patient complaints, media stories<br />
<strong>and</strong> other sources. 29 These statistics are clearly affected by underreporting <strong>and</strong> consist primarily<br />
<strong>of</strong> serious adverse events (76% <strong>of</strong> events reported resulted in patient deaths), not near misses. As<br />
discussed in the chapter on wrong-site surgeries (Subchapter 43.2), comparing JCAHO reports<br />
with data from the m<strong>and</strong>atory incident reporting system maintained by the New York State<br />
Department <strong>of</strong> Health 36 suggests that the JCAHO statistics underestimate the true incidence <strong>of</strong><br />
target events by at least a factor <strong>of</strong> 20.<br />
43
Opportunities for Impact<br />
Most hospitals’ incident reporting systems fail to capture the majority <strong>of</strong> errors <strong>and</strong> near<br />
misses. 24 Studies <strong>of</strong> medical services suggest that only 1.5% <strong>of</strong> all adverse events result in an<br />
incident report 37 <strong>and</strong> only 6% <strong>of</strong> adverse drug events are identified through traditional incident<br />
reporting or a telephone hotline. 24 The American College <strong>of</strong> Surgeons estimates that incident<br />
reports generally capture only 5-30% <strong>of</strong> adverse events. 38 A study <strong>of</strong> a general surgery service<br />
showed that only 20% <strong>of</strong> complications on a surgical service ever resulted in discussion at<br />
Morbidity <strong>and</strong> Mortality rounds. 39 Given the endemic underreporting revealed in the literature,<br />
modifications to the configuration <strong>and</strong> operation <strong>of</strong> the typical hospital reporting system could<br />
yield higher capture rates <strong>of</strong> relevant clinical data.<br />
Study Designs<br />
We analyzed 5 studies that evaluated different methods <strong>of</strong> critical incident reporting.<br />
Two studies prospectively investigated incident reporting compared with observational data<br />
collection 24,39 <strong>and</strong> one utilized retrospective chart review. 37 Two additional studies looked at<br />
enhanced incident reporting by active solicitation <strong>of</strong> physician input compared with background<br />
hospital quality assurance (QA) measures. 40,41 In addition, we reviewed JCAHO’s report <strong>of</strong> its<br />
Sentinel Event Database, <strong>and</strong> the Australian Incident Monitoring Study, both because <strong>of</strong> the<br />
large sizes <strong>and</strong> the high pr<strong>of</strong>iles <strong>of</strong> the studies. 13,26 Additional reports <strong>of</strong> critical incident<br />
reporting systems in the medical literature consist primarily <strong>of</strong> uncontrolled observational<br />
trials 42-44 that are not reviewed in this chapter.<br />
Study Outcomes<br />
In general, published studies <strong>of</strong> incident reporting do not seek to establish the benefit <strong>of</strong><br />
incident reporting as a patient safety practice. Their principal goal is to determine if incident<br />
reporting, as it is practiced, captures the relevant events. 40 In fact, no studies have established the<br />
value <strong>of</strong> incident reporting on patient safety outcomes.<br />
The large JCAHO <strong>and</strong> Australian databases provide data about reporting rates, <strong>and</strong> an<br />
array <strong>of</strong> quantitative <strong>and</strong> qualitative information about the reported incidents, including the<br />
identity <strong>of</strong> the reporter, time <strong>of</strong> report, severity <strong>and</strong> type <strong>of</strong> error. 13,26 Clearly these do not<br />
represent clinical outcomes, but they may be reasonable surrogates for the organizational focus<br />
on patient safety. For instance, increased incident reporting rates may not be indicative <strong>of</strong> an<br />
unsafe organization, 45 but may reflect a shift in organizational culture to increased acceptance <strong>of</strong><br />
quality improvement <strong>and</strong> other organizational changes. 3,5<br />
None <strong>of</strong> the studies reviewed captured outcomes such as morbidity or error rates. The<br />
AIMS group published an entire symposium which reported the quantitative <strong>and</strong> qualitative data<br />
regarding 2000 critical incidents in anesthesia. 13 However, only a small portion <strong>of</strong> these<br />
incidents were prospectively evaluated. 14 None <strong>of</strong> the studies reviewed for this chapter<br />
performed formal root cause analyses on reported errors (Chapter 5).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
As described above, 6 years <strong>of</strong> JCAHO sentinel event data have captured merely 1152<br />
events, none <strong>of</strong> which include near misses. 29 Despite collecting what is likely to represent only a<br />
fraction <strong>of</strong> the target events, JCAHO has compiled the events, reviewed the root cause analyses<br />
<strong>and</strong> provided recommendations for procedures to improve patient safety for events ranging from<br />
wrong-site surgeries to infusion pump-related adverse events (Table 4.2). This information may<br />
44
prove to be particularly useful in the case <strong>of</strong> rare events such as wrong-site surgery, where<br />
national collection <strong>of</strong> incidents can yield a more statistically useful sample size.<br />
The first 2000 incident reports to AIMS from 90 member institutions were published in<br />
1993. 13 In contrast to the JCAHO data, all events were self-reported by anesthetists <strong>and</strong> only 2%<br />
<strong>of</strong> events reported resulted in patient deaths. A full 44% <strong>of</strong> events had negligible effect on<br />
patient outcome. Ninety percent <strong>of</strong> reports had identified systems failures, <strong>and</strong> 79% had<br />
identified human failures. The AIMS data were similar to those <strong>of</strong> Cooper 19 in terms <strong>of</strong> percent<br />
<strong>of</strong> incidents with reported human failures, timing <strong>of</strong> events with regard to phase <strong>of</strong> anesthesia,<br />
<strong>and</strong> type <strong>of</strong> events (breathing circuit misconnections were between 2% <strong>and</strong> 3% in both<br />
studies). 13,19 The AIMS data are also similar to American “closed-claims” data in terms <strong>of</strong><br />
pattern, nature <strong>and</strong> proportion <strong>of</strong> the total number <strong>of</strong> reports for several types <strong>of</strong> adverse events, 13<br />
which lends further credibility to the reports.<br />
The AIMS data, although also likely to be affected by underreporting because <strong>of</strong> its<br />
voluntary nature, clearly captures a higher proportion <strong>of</strong> critical incidents than the JCAHO<br />
Sentinel Event Database. Despite coming from only 90 participating sites, AIMS received more<br />
reports over a similar time frame than the JCAHO did from the several thous<strong>and</strong> accredited<br />
United States hospitals. This disparity may be explained by the fact that AIMS institutions were<br />
self-selected, <strong>and</strong> that the culture <strong>of</strong> anesthesia is more attuned to patient safety concerns. 47<br />
The poor capture rate <strong>of</strong> incident reporting systems in American hospitals has not gone<br />
unnoticed. Cullen et al 24 prospectively investigated usual hospital incident reporting compared to<br />
observational data collection for adverse drug events (ADE logs, daily solicitation from hospital<br />
personnel, <strong>and</strong> chart review) in 5 patient care units <strong>of</strong> a tertiary care hospital. Only 6% <strong>of</strong> ADEs<br />
were identified <strong>and</strong> only 8% <strong>of</strong> serious ADEs were reported. These findings are similar to those<br />
in the pharmacy literature, 48,49 <strong>and</strong> are attributed to cultural <strong>and</strong> environmental factors. A similar<br />
study on a general surgical service found that 40% <strong>of</strong> patients suffered complications. 39 While<br />
chart documentation was excellent (94%), only 20% <strong>of</strong> complications were discussed at<br />
Morbidity <strong>and</strong> Mortality rounds.<br />
Active solicitation <strong>of</strong> physician reporting has been suggested as a way to improve<br />
adverse event <strong>and</strong> near miss detection rates. Weingart et al 41 employed direct physician<br />
interviews supplemented by email reminders to increase detection <strong>of</strong> adverse events in a tertiary<br />
care hospital. The physicians reported an entirely unique set <strong>of</strong> adverse events compared with<br />
those captured by the hospital incident reporting system. Of 168 events, only one was reported<br />
by both methods. O’Neil et al 37 used e-mail to elicit adverse events from housestaff <strong>and</strong><br />
compared these with those found on retrospective chart review. Of 174 events identified, 41<br />
were detected by both methods. The house <strong>of</strong>ficers appeared to capture preventable adverse<br />
events at a higher rate (62.5% v. 32%, p=0.003). In addition, the hospital’s risk management<br />
system detected only 4 <strong>of</strong> 174 adverse events. Welsh et al 40 employed prompting <strong>of</strong> house<br />
<strong>of</strong>ficers at morning report to augment hospital incident reporting systems. There was overlap in<br />
reporting in only 2.6% <strong>of</strong> 341 adverse events that occurred during the study. In addition,<br />
although the number <strong>of</strong> events house <strong>of</strong>ficers reported increased with daily prompting, the<br />
quantity rapidly decreased when prompting ceased. In summary, there is evidence that active<br />
solicitation <strong>of</strong> critical incident reports by physicians can augment existing databases, identifying<br />
incidents not detected through other means, although the response may not be durable. 37,40,41<br />
Potential for Harm<br />
Users may view reporting systems with skepticism, particularly the system’s ability to<br />
maintain confidentiality <strong>and</strong> shield participants from legal exposure. 28 In many states, critical<br />
45
incident reporting <strong>and</strong> analysis count as peer review activities <strong>and</strong> are protected from legal<br />
discovery. 28,50 However, other states <strong>of</strong>fer little or no protection, <strong>and</strong> reporting events to external<br />
agencies (eg, to JCAHO) may obliterate the few protections that do exist. In recognition <strong>of</strong> this<br />
problem, JCAHO’s Terms <strong>of</strong> Agreement with hospitals now includes a provision identifying<br />
JCAHO as a participant in each hospital’s quality improvement process. 28<br />
Costs <strong>and</strong> Implementation<br />
Few estimates <strong>of</strong> costs have been reported in the literature. In general, authors remarked<br />
that incident reporting was far less expensive than retrospective review. One single center study<br />
estimated that physician reporting was less costly ($15,000) than retrospective record review<br />
($54,000) over a 4-month period. 37 A survey <strong>of</strong> administrators <strong>of</strong> reporting systems from nonmedical<br />
industries reported a consensus that costs were far <strong>of</strong>fset by the potential benefits. 6<br />
Comment<br />
The wide variation in reporting <strong>of</strong> incidents may have more to do with reporting<br />
incentives <strong>and</strong> local culture than with the quality <strong>of</strong> medicine practiced there. 24 When institutions<br />
prioritize incident reporting among medical staff <strong>and</strong> trainees, however, the incident reporting<br />
systems seem to capture a distinct set <strong>of</strong> events from those captured by chart review <strong>and</strong><br />
traditional risk management 40,41 <strong>and</strong> events captured in this manner may be more preventable. 37<br />
The addition <strong>of</strong> anonymous or non-punitive systems is likely to increase the rates <strong>of</strong><br />
incident reporting <strong>and</strong> detection. 51 Other investigators have also noted increases in reporting<br />
when new systems are implemented <strong>and</strong> a culture conducive to reporting is maintained. 40,52<br />
Several studies suggest that direct solicitation <strong>of</strong> physicians results in reports that are more likely<br />
to be distinct, preventable, <strong>and</strong> more severe than those obtained by other means. 8,37,41<br />
The nature <strong>of</strong> incident reporting, replete with hindsight bias, lost information, <strong>and</strong> lost<br />
contextual clues makes it unlikely that robust data will ever link it directly with improved<br />
outcomes. Nonetheless, incident reporting appears to be growing in importance in medicine. The<br />
Institute <strong>of</strong> Medicine report, To Err is Human, 53 has prompted calls for m<strong>and</strong>atory reporting <strong>of</strong><br />
medical errors to continue in the United States. 54-57 Engl<strong>and</strong>’s National Health Service plans to<br />
launch a national incident reporting system as well, which has raised concerns similar to those<br />
voiced in the American medical community. 58 While the literature to date does not permit an<br />
evidence-based resolution <strong>of</strong> the debate over m<strong>and</strong>atory versus voluntary incident reporting, it is<br />
clear that incident reporting represents just one <strong>of</strong> several potential sources <strong>of</strong> information about<br />
patient safety <strong>and</strong> that these sources should be regarded as complementary. In other industries<br />
incident reporting has succeeded when it is m<strong>and</strong>ated by regulatory agencies or is anonymous<br />
<strong>and</strong> voluntary on the part <strong>of</strong> reporters, <strong>and</strong> when it provides incentives <strong>and</strong> feedback to<br />
reporters. 6 The ability <strong>of</strong> health care organizations to replicate the successes <strong>of</strong> other industries in<br />
their use <strong>of</strong> incident reporting systems 6 will undoubtedly depend in large part on the uses to<br />
which they put these data. Specifically, success or failure may depend on whether health care<br />
organizations use the data to fuel institutional quality improvement rather than to generate<br />
individual performance evaluations.<br />
46
Table 4.1. Examples <strong>of</strong> events reported to hospital incident reporting systems 25-27<br />
Adverse Outcomes Procedural Breakdowns Catastrophic Events<br />
Unexpected death or<br />
disability<br />
Inpatient falls or<br />
“mishaps”<br />
Institutionally-acquired<br />
burns<br />
Institutionally-acquired<br />
pressure sores<br />
Errors or unexpected<br />
complications related to the<br />
administration <strong>of</strong> drugs or<br />
transfusion<br />
Discharges against medical<br />
advice (“AMA”)<br />
Significant delays in<br />
diagnosis or diagnostic<br />
testing<br />
Breach <strong>of</strong> confidentiality<br />
47<br />
Performance <strong>of</strong> a procedure on the<br />
wrong body part (“wrong-site<br />
surgery”)<br />
Performance <strong>of</strong> a procedure on the<br />
wrong patient<br />
Infant abduction or discharge to<br />
wrong family<br />
Rape <strong>of</strong> a hospitalized patient<br />
Suicide <strong>of</strong> a hospitalized patient<br />
Table 4.2. Sentinel event alerts published by JCAHO following analysis <strong>of</strong> incident<br />
reports 46<br />
• Medication Error Prevention – Potassium Chloride<br />
• Lessons Learned: Wrong Site Surgery<br />
• Inpatient Suicides: Recommendations for Prevention<br />
• Preventing Restraint Deaths<br />
• Infant Abductions: Preventing Future Occurrences<br />
• High-Alert Medications <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong><br />
• Operative <strong>and</strong> Postoperative Complications: Lessons for the Future<br />
• Fatal Falls: Lessons for the Future<br />
• Infusion Pumps: Preventing Future Adverse Events<br />
• Lessons Learned: Fires in the Home Care Setting<br />
• Kernicterus Threatens Healthy Newborns
References<br />
1. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, et al. Incidence<br />
<strong>of</strong> adverse events <strong>and</strong> negligence in hospitalized patients. Results <strong>of</strong> the Harvard <strong>Medical</strong><br />
Practice Study I. N Engl J Med. 1991;324:370-376.<br />
2. Studdert D, Thomas E, Burstin H, Zbar B, Orav J, Brennan T. Negligent care <strong>and</strong><br />
malpractice claiming behavior in Utah <strong>and</strong> Colorado. <strong>Medical</strong> Care. 2000;38:250-260.<br />
3. Van der Schaaf T. Near miss reporting in the Chemical Process Industry [Doctoral Thesis].<br />
Eindhoven, The Netherl<strong>and</strong>s: Eindhoven University <strong>of</strong> Technology; 1992.<br />
4. Ibojie J, Urbaniak S. Comparing near misses with actual mistransfusion events a more<br />
accurate reflection <strong>of</strong> transfusion errors. Br J Haematol. 2000;108:458-460.<br />
5. Battles J, Kaplan H, Van der Schaaf T, Shea C. The attributes <strong>of</strong> medical event-reporting<br />
systems. Arch Pathol Lab Med. 1998;122:231-238.<br />
6. Barach P, Small S. Reporting <strong>and</strong> preventing medical mishaps: lessons from non-medical<br />
near miss reporting systems. BMJ. 2000;320:759-763.<br />
7. Whipple J, Quebbeman E, Lewis K, Gaughan L, Gallup E, Ausman R. Identification <strong>of</strong><br />
patient controlled analgesia overdoses in hospitalized patients: a computerized method <strong>of</strong><br />
monitoring adverse events. Ann Pharmacother. 1994;28:655-658.<br />
8. Bates D, Makary M, Teich J, Pedraza L, Ma'luf N, Burstin H, et al. Asking residents about<br />
adverse events in a computer dialogue: how accurate are they? Jt Comm J Qual Improv.<br />
1998;24:197-202.<br />
9. Jha A, GJ K, Teich J. Identifying adverse drug events: development <strong>of</strong> a computer-based<br />
monitor <strong>and</strong> comparison with chart review <strong>and</strong> stimulated voluntary report. J Am Med<br />
Inform Assoc. 1998;5:305-314.<br />
10. Naranjo C, Lanctot K. Recent developments in computer-assisted diagnosis <strong>of</strong> putative<br />
adverse drug reactions. Drug Saf. 1991;6:315-322.<br />
11. Classen D, Pestotnik S, Evans R, Burke J. Computerized surveillance <strong>of</strong> adverse drug<br />
events in hospital patients. JAMA. 1991;266:2847-2851.<br />
12. Flanagan J. The critical incident technique. Psychol Bull. 1954;51:327-358.<br />
13. Webb R, Currie M, Morgan C, Williamson J, Mackay P, Russell W, et al. The Australian<br />
Incident Monitoring Study: an analysis <strong>of</strong> 2000 incident reports. Anaesth Intens Care.<br />
1993;21:520-528.<br />
14. Runciman W, Sellen A, Webb R, Williamson J, Currie M, Morgan C, et al. Errors,<br />
incidents <strong>and</strong> accidents in anaesthetic practice. Anaesth Intens Care. 1993;21:506-519.<br />
15. Fisch<strong>of</strong>f B. Hindsight does not equal foresight: the effect <strong>of</strong> outcome knowledge on<br />
judgment under uncertainty. J Exp Psych: Human Perform <strong>and</strong> Percept. 1975;1:288-299.<br />
16. Food <strong>and</strong> Drug Administration: Biological products; reporting <strong>of</strong> errors <strong>and</strong> accidents in<br />
manufacturing. Federal Register. 1997;62:49642-49648.<br />
17. Safren MA, Chapanis A. A critical incident study <strong>of</strong> hospital medication errors. Hospitals.<br />
1960;34:32-34.<br />
18. Graham JR, Winslow WW, Hickey MA. The critical incident technique applied to<br />
postgraduate training in psychiatry. Can Psychiatr Assoc J. 1972;17:177-181.<br />
19. Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study<br />
<strong>of</strong> human factors. Anesthesiology. 1978;49:399-406.<br />
20. Buckley T, Short T, Rowbottom Y, OH T. <strong>Critical</strong> incident reporting in the intensive care<br />
unit. Anaesthesia. 1997;52:403-409.<br />
48
21. Hart G, Baldwin I, Gutteridge G, Ford J. Adverse incident reporting in intensive care.<br />
Anaesth Intens Care. 1994;22:556-561.<br />
22. Linden JV, Paul B, Dressler KP. A report <strong>of</strong> 104 transfusion errors in New York State.<br />
Transfusion. 1992;32:601-606.<br />
23. Emori T, Edwards J, Culver D, Sartor C, Stroud L, Gaunt E, et al. Accuracy <strong>of</strong> reporting<br />
nosocomial infections in intensive-care-unit patients to the National Nosocomial Infections<br />
Surveillance System: a pilot study. Infect Control Hosp Epidemiol. 1998;19:308-316.<br />
24. Cullen D, Bates D, Small S, Cooper J, Nemeskal A, Leape L. The incident reporting<br />
system does not detect adverse events: a problem for quality improvement. Jt Comm J<br />
Qual Improv. 1995;21:541-548.<br />
25. Incident reports. A guide to legal issues in health care. Philadelphia, PA: Trustees <strong>of</strong> the<br />
University <strong>of</strong> Pennsylvania. 1998:127-129.<br />
26. Joint Commission on the Accreditation <strong>of</strong> Healthcare Organizations. Sentinel event policy<br />
<strong>and</strong> procedures. Available at: http://JCAHO.org/sentinel/se_pp.html. Accessed March 15,<br />
2001.<br />
27. Fischer G, Fetters M, Munro A, Goldman E. Adverse events in primary care identified<br />
from a risk-management database. J Fam Pract. 1997;45:40-46.<br />
28. Berman S. Identifying <strong>and</strong> addressing sentinel events: an Interview with Richard Croteau.<br />
Jt Comm J Qual Improv. 1998;24:426-434.<br />
29. Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Statistics.<br />
Available at: http://www.JCAHO.org/sentinel/se_stats.html. Accessed April 16, 2001.<br />
30. Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Alert.<br />
Available at: http://JCAHO.org/edu_pub/sealert/se_alert.html. Accessed May 14, 2001.<br />
31. Lessons learned: sentinel event trends in wrong-site surgery. Joint Commission<br />
Perspectives. 2000;20:14.<br />
32. Symposium: The Australian Incident Monitoring Study. Anaesth Intens Care.<br />
1993;21:506-695.<br />
33. Billings C, Reynard W. Human factors in aircraft incidents: results <strong>of</strong> a 7-year study. Aviat<br />
Space Environ Med. 1984;55:960-965.<br />
34. National Aeronautics & Space Agency (NASA). Aviation <strong>Safety</strong> Reporting System.<br />
Available at: http://www-afo.arc.nasa.gov/ASRS/ASRS.html. Accessed March 14, 2000.<br />
35. Leape L. Reporting <strong>of</strong> medical errors: time for a reality check. Qual Health Care.<br />
2000;9:144-145.<br />
36. New York State Department <strong>of</strong> Health. NYPORTS - The New York <strong>Patient</strong> Occurrence<br />
<strong>and</strong> Tracking System. Available at:<br />
http://www.health.state.ny.us/nysdoh/commish/2001/nyports/nyports.htm. Accessed May<br />
31, 2001.<br />
37. O'Neil A, Petersen L, Cook E, Bates D, Lee T, Brennan T. Physician reporting compared<br />
with medical-record review to identify adverse medical events. Ann Intern Med.<br />
1993;119:370-376.<br />
38. Data sources <strong>and</strong> coordination. In: Surgeons ACo, editor. <strong>Patient</strong> safety manual. Rockville,<br />
MD: Bader & Associates, Inc. 1985.<br />
39. Wanzel K, Jamieson C, Bohnen J. Complications on a general surgery service: incidence<br />
<strong>and</strong> reporting. CJS. 2000;43:113-117.<br />
40. Welsh C, Pedot R, Anderson R. Use <strong>of</strong> morning report to enhance adverse event detection.<br />
J Gen Intern Med. 1996;11:454-460.<br />
49
41. Weingart SN, Ship AN, Aronson MD. Confidential clinician-reported surveillance <strong>of</strong><br />
adverse events among medical inpatients. J Gen Intern Med. 2000;15:470-477.<br />
42. Frey B, Kehrer B, Losa M, Braun H, Berweger L, Micallef J, et al. Comprehensive critical<br />
incident monitoring in a neonatal-pediatric intensive care unit: experience with the system<br />
approach. Intensive Care Med. 2000;26:69-74.<br />
43. Findlay G, Spittal M, Radcliffe J. The recognition <strong>of</strong> critical incidents: quantification <strong>of</strong><br />
monitor effectiveness. Anaesthesia. 1998;53:589-603.<br />
44. Flaatten H, Hevroy O. Errors in the intensive care unit (ICU). Act Anaesthesiol Sc<strong>and</strong>.<br />
1999;43:614-617.<br />
45. Connolly C. Reducing error, improving safety: relation between reported mishaps <strong>and</strong><br />
safety is unclear. BMJ. 2000;321:505.<br />
46. The Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Alert.<br />
Available at: http://www.JCAHO.org/edu_pub/sealert/se_alert.html. Accessed May 31,<br />
2001.<br />
47. Gaba D. Anaesthesiology as a model for patient safety in health care. BMJ. 2000;320.<br />
48. Shannon R, De Muth J. Comparison <strong>of</strong> medication error detection methods in the long term<br />
care facility. Consulting Pharmacy. 1987;2:148-151.<br />
49. Allan E, Barker K. Fundamentals <strong>of</strong> medication error research. Am J Hosp Pharm.<br />
1990;47:555-571.<br />
50. Rex J, Turnbull J, Allen S, V<strong>and</strong>e Voorde K, Luther K. Systematic root cause analysis <strong>of</strong><br />
adverse drug events in a tertiary referral hospital. Jt Comm J Qual Improv. 2000;26:563-<br />
575.<br />
51. Kaplan H, Battles J, Van Der Schaaf T, Shea C, Mercer S. Identification <strong>and</strong> classification<br />
<strong>of</strong> the causes <strong>of</strong> events in transfusion medicine. Transfusion. 1998;38:1071-1081.<br />
52. Hartwig S, Denger S, Schneider P. Severity-indexed, incident report-based medication<br />
error-reporting program. Am J Hosp Pharm. 1991;48:2611-2616.<br />
53. Kohn L, Corrigan J, Donaldson M, editors. To Err Is Human: Building a Safer Health<br />
System. Washington, DC: Committee on Quality <strong>of</strong> Health Care in America, Institute <strong>of</strong><br />
Medicine. National Academy Press. 2000.<br />
54. Pear R. U.S. health <strong>of</strong>ficials reject plan to report medical mistakes. New York Times Jan 24<br />
2000: A14(N), A14(L).<br />
55. Prager LO. M<strong>and</strong>atory reports cloud error plan: supporters are concerned that the merits <strong>of</strong><br />
a plan to reduce medical errors by 50% are being obscured by debate over its most<br />
controversial component. American <strong>Medical</strong> News March 13 2000:1,66-67.<br />
56. Murray S. Clinton to call on all states to adopt systems for reporting medical errors. Wall<br />
Street Journal Feb 22 2000:A28(W), A6(E).<br />
57. Kaufman M. Clinton Seeks <strong>Medical</strong> Error Reports; Proposal to Reduce Mistakes Includes<br />
M<strong>and</strong>atory Disclosure, Lawsuit Shield. Washington Post Feb 22 2000:A02.<br />
58. Mayor S. English NHS to set up new reporting system for errors. BMJ. 2000;320:1689.<br />
50
Chapter 5. Root Cause <strong>Analysis</strong><br />
Heidi Wald, MD<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Historically, medicine has relied heavily on quantitative approaches for quality<br />
improvement <strong>and</strong> error reduction. For instance, the US Food <strong>and</strong> Drug Administration (FDA)<br />
has collected data on major transfusion errors since the mid-1970s. 1,2 Using the statistical power<br />
<strong>of</strong> these nationwide data, the most common types <strong>of</strong> errors have been periodically reviewed <strong>and</strong><br />
systems improvements recommended. 3<br />
These epidemiologic techniques are suited to complications that occur with reasonable<br />
frequency, but not for rare (but nonetheless important) errors. Outside <strong>of</strong> medicine, high-risk<br />
industries have developed techniques to address major accidents. Clearly the nuclear power<br />
industry cannot wait for several Three Mile Isl<strong>and</strong>-type events to occur in order to conduct valid<br />
analyses to determine the likely causes.<br />
A retrospective approach to error analysis, called root cause analysis (RCA), is widely<br />
applied to investigate major industrial accidents. 4 RCA has its foundations in industrial<br />
psychology <strong>and</strong> human factors engineering. Many experts have championed it for the<br />
investigation <strong>of</strong> sentinel events in medicine. 5-7 In 1997, the Joint Commission on the<br />
Accreditation <strong>of</strong> Healthcare Organizations (JCAHO) m<strong>and</strong>ated the use <strong>of</strong> RCA in the<br />
investigation <strong>of</strong> sentinel events in accredited hospitals. 8<br />
The most commonly cited taxonomy <strong>of</strong> human error in the medical literature is based on<br />
the work <strong>of</strong> James Reason. 4,9,10 Reason describes 2 major categories <strong>of</strong> error: active error, which<br />
generally occurs at the point <strong>of</strong> human interface with a complex system, <strong>and</strong> latent error, which<br />
represents failures <strong>of</strong> system design. RCA is generally employed to uncover latent errors<br />
underlying a sentinel event. 6,7<br />
RCA provides a structured <strong>and</strong> process-focused framework with which to approach<br />
sentinel event analysis. Its cardinal tenet is to avoid the pervasive <strong>and</strong> counterproductive culture<br />
<strong>of</strong> individual blame. 11,12 Systems <strong>and</strong> organizational issues can be identified <strong>and</strong> addressed, <strong>and</strong><br />
active errors are acknowledged. 6 Systematic application <strong>of</strong> RCA may uncover common root<br />
causes that link a disparate collection <strong>of</strong> accidents (ie, a variety <strong>of</strong> serious adverse events<br />
occurring at shift change). Careful analysis may suggest system changes designed to prevent<br />
future incidents. 13<br />
Despite these intriguing qualities, RCA has significant methodologic limitations. RCAs<br />
are in essence uncontrolled case studies. As the occurrence <strong>of</strong> accidents is highly unpredictable,<br />
it is impossible to know if the root cause established by the analysis is the cause <strong>of</strong> the<br />
accident. 14 In addition, RCAs may be tainted by hindsight bias. 4,15,16 Other biases stem from how<br />
deeply the causes are probed <strong>and</strong> influenced by the prevailing concerns <strong>of</strong> the day. 16,17 The fact<br />
that technological failures (device malfunction), which previously represented the focus <strong>of</strong> most<br />
accident analyses, have been supplanted by staffing issues, management failures, <strong>and</strong><br />
information systems problems may be an example <strong>of</strong> the latter bias. 17 Finally, RCAs are timeconsuming<br />
<strong>and</strong> labor intensive.<br />
51
Despite legitimate concerns about the place <strong>of</strong> RCA in medical error reduction, the<br />
JCAHO m<strong>and</strong>ate ensures that RCA will be widely used to analyze sentinel events. 8 Qualitative<br />
methods such as RCA should be used to supplement quantitative methods, to generate new<br />
hypotheses, <strong>and</strong> to examine events not amenable to quantitative methods (for example, those that<br />
occur rarely). 18 As such, its credibility as a research tool should be judged by the st<strong>and</strong>ards<br />
appropriate for qualitative research, not quantitative. 19,20 Yet, the outcomes <strong>and</strong> costs associated<br />
with RCA are largely unreported. This chapter reviews the small body <strong>of</strong> published literature<br />
regarding the use <strong>of</strong> RCA in the investigation <strong>of</strong> medical errors.<br />
Practice Description<br />
To be credible, RCA requires rigorous application <strong>of</strong> established qualitative techniques.<br />
Once a sentinel event has been identified for analysis (eg, a major chemotherapy dosing error, a<br />
case <strong>of</strong> wrong-site surgery, or major ABO incompatible transfusion reaction), a multidisciplinary<br />
team is assembled to direct the investigation. The members <strong>of</strong> this team should be trained in the<br />
techniques <strong>and</strong> goals <strong>of</strong> RCA, as the tendency to revert to personal biases is strong. 13,14 Multiple<br />
investigators allow triangulation or corroboration <strong>of</strong> major findings <strong>and</strong> increase the validity <strong>of</strong><br />
the final results. 19 Based on the concepts <strong>of</strong> active <strong>and</strong> latent error described above, accident<br />
analysis is generally broken down into the following steps 6,7 :<br />
1. Data collection: establishment <strong>of</strong> what happened through structured<br />
interviews, document review, <strong>and</strong>/or field observation. These data are used to<br />
generate a sequence or timeline <strong>of</strong> events preceeding <strong>and</strong> following the event;<br />
2. Data analysis: an iterative process to examine the sequence <strong>of</strong> events<br />
generated above with the goals <strong>of</strong> determining the common underlying<br />
factors:<br />
(i) Establishment <strong>of</strong> how the event happened by identification <strong>of</strong> active<br />
failures in the sequence;<br />
(ii) Establishment <strong>of</strong> why the event happened through identification <strong>of</strong> latent<br />
failures in the sequence which are generalizable.<br />
In order to ensure consideration <strong>of</strong> all potential root causes <strong>of</strong> error, one popular<br />
conceptual framework for contributing factors has been proposed based on work by Reason. 7<br />
Several other frameworks also exist. 21,22 The categories <strong>of</strong> factors influencing clinical practice<br />
include institutional/regulatory, organizational/management, work environment, team factors,<br />
staff factors, task factors, <strong>and</strong> patient characteristics. Each category can be exp<strong>and</strong>ed to provide<br />
more detail. A credible RCA considers root causes in all categories before rejecting a factor or<br />
category <strong>of</strong> factors as non-contributory. A st<strong>and</strong>ardized template in the form <strong>of</strong> a tree (or<br />
“Ishikawa”) may help direct the process <strong>of</strong> identifying contributing factors, with such factors<br />
leading to the event grouped (on tree “roots”) by category. Category labels may vary depending<br />
on the setting. 23<br />
At the conclusion <strong>of</strong> the RCA, the team summarizes the underlying causes <strong>and</strong> their<br />
relative contributions, <strong>and</strong> begins to identify administrative <strong>and</strong> systems problems that might be<br />
c<strong>and</strong>idates for redesign. 6<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
JCAHO’s 6-year-old sentinel event database <strong>of</strong> voluntarily reported incidents (see<br />
Chapter 4) has captured a mere 1152 events, <strong>of</strong> which 62% occurred in general hospitals. Twothirds<br />
<strong>of</strong> the events were self-reported by institutions, with the remainder coming from patient<br />
52
complaints, media stories <strong>and</strong> other sources. 24 These statistics are clearly affected by<br />
underreporting <strong>and</strong> consist primarily <strong>of</strong> serious adverse events (76% <strong>of</strong> events reported resulted<br />
in patient deaths), not near misses. The number <strong>of</strong> sentinel events appropriate for RCA is likely<br />
to be orders <strong>of</strong> magnitude greater.<br />
The selection <strong>of</strong> events for RCA may be crucial to its successful implementation on a<br />
regular basis. Clearly, it cannot be performed for every medical error. JCAHO provides guidance<br />
for hospitals about which events are considered “sentinel,” 8 but the decision to conduct RCA is<br />
at the discretion <strong>of</strong> the leadership <strong>of</strong> the organization. 12<br />
If the number <strong>of</strong> events is large <strong>and</strong> homogeneous, many events can be excluded from<br />
analysis. In a transfusion medicine reporting system, all events were screened after initial report<br />
<strong>and</strong> entered in the database, but those not considered sufficiently unique did not undergo RCA. 25<br />
Opportunities for Impact<br />
While routine RCA <strong>of</strong> sentinel events is m<strong>and</strong>ated, the degree to which hospitals carry<br />
out credible RCAs is unknown. Given the numerous dem<strong>and</strong>s on hospital administrators <strong>and</strong><br />
clinical staff, it is likely that many hospitals fail to give this process a high pr<strong>of</strong>ile, assigning the<br />
task to a few personnel with minimal training in RCA rather than involving trained leaders from<br />
all relevant departments. The degree <strong>of</strong> underreporting to JCAHO suggests that many hospitals<br />
are wary <strong>of</strong> probationary status <strong>and</strong> the legal implications <strong>of</strong> disclosure <strong>of</strong> sentinel events <strong>and</strong> the<br />
results <strong>of</strong> RCAs. 12,26<br />
Study Designs<br />
As RCA is a qualitative technique, most reports in the literature are case studies or case<br />
series <strong>of</strong> its application in medicine. 6,27-30 There is little published literature that systematically<br />
evaluates the impact <strong>of</strong> formal RCA on error rates. The most rigorous study comes from a<br />
tertiary referral hospital in Texas that systematically applied RCA to all serious adverse drug<br />
events (ADEs) considered preventable. The time series contained background data during the<br />
initial implementation period <strong>of</strong> 12 months <strong>and</strong> a 17-month follow-up phase. 13<br />
Study Outcomes<br />
Published reports <strong>of</strong> the application <strong>of</strong> RCA in medicine generally present incident<br />
reporting rates, categories <strong>of</strong> active errors determined by the RCA, categories <strong>of</strong> root causes<br />
(latent errors) <strong>of</strong> the events, <strong>and</strong> suggested systems improvements. While these do not represent<br />
clinical outcomes, they are reasonable surrogates for evaluation. For instance, increased incident<br />
reporting rates may reflect an institution’s shift toward increased acceptance <strong>of</strong> quality<br />
improvement <strong>and</strong> organizational change. 5,21<br />
53
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The Texas study revealed a 45% decrease in the rate <strong>of</strong> voluntarily reported serious<br />
ADEs between the study <strong>and</strong> follow-up periods (7.2 per 100,000 to 4.0 per 100,000 patient-days,<br />
p
investigation in the name <strong>of</strong> quality improvement, with one health system reporting a sustained<br />
10-fold increase in reporting. 25,27<br />
Comment<br />
Root cause analyses systematically search out latent or system failures that underlie<br />
adverse events or near misses. They are limited by their retrospective <strong>and</strong> inherently speculative<br />
nature. There is insufficient evidence in the medical literature to support RCA as a proven<br />
patient safety practice, however it may represent an important qualitative tool that is<br />
complementary to other techniques employed in error reduction. When applied appropriately,<br />
RCA may illuminate targets for change, <strong>and</strong>, in certain health care contexts, may generate<br />
testable hypotheses. The use <strong>of</strong> RCA merits more consideration, as it lends a formal structure to<br />
efforts to learn from past mistakes.<br />
References<br />
1. Sazama K. Current good manufacturing practices for transfusion medicine. Transfus Med<br />
Rev. 1996;10:286-295.<br />
2. Food <strong>and</strong> Drug Administration: Biological products; reporting <strong>of</strong> errors <strong>and</strong> accidents in<br />
manufacturing. Fed Regist. 1997;62:49642-49648.<br />
3. Sazama K. Reports <strong>of</strong> 355 transfusion-associated deaths: 1976 through 1985. Transfusion.<br />
1990;30:583-590.<br />
4. Reason JT. Human Error. New York: Cambridge Univ Press. 1990.<br />
5. Battles JB, Kaplan HS, Van der Schaaf TW, Shea CE. The attributes <strong>of</strong> medical eventreporting<br />
systems: experience with a prototype medical event-reporting system for<br />
transfusion medicine. Arch Pathol Lab Med. 1998;122:231-238.<br />
6. Eagle CJ, Davies JM, Reason J. Accident analysis <strong>of</strong> large-scale technological disasters<br />
applied to an anaesthetic complication. Can J Anaesth. 1992;39:118-122.<br />
7. Vincent C, Ennis M, Audley RJ. <strong>Medical</strong> accidents. Oxford ; New York: Oxford<br />
University Press. 1993.<br />
8. Joint Commission on the Accreditation <strong>of</strong> Healthcare Organizations. Sentinel event policy<br />
<strong>and</strong> procedures. Available at: http://www.JCAHO.org/sentinel/se_pp.html. Accessed May<br />
30, 2001.<br />
9. Reason JT. Managing the Risks <strong>of</strong> Organizational Accidents: Ashgate Publishing<br />
Company. 1997.<br />
10. Reason J. Human error: models <strong>and</strong> management. BMJ. 2000;320:768-770.<br />
11. Leape LL. Error in medicine. JAMA. 1994;272:1851-1857.<br />
12. Berman S. Identifying <strong>and</strong> addressing sentinel events: an Interview with Richard Croteau.<br />
Jt Comm J Qual Improv. 1998;24:426-434.<br />
13. Rex JH, Turnbull JE, Allen SJ, V<strong>and</strong>e Voorde K, Luther K. Systematic root cause analysis<br />
<strong>of</strong> adverse drug events in a tertiary referral hospital. Jt Comm J Qual Improv. 2000;26:563-<br />
575.<br />
14. Runciman WB, Sellen A, Webb RK, Williamson JA, Currie M, Morgan C, et al. The<br />
Australian Incident Monitoring Study. Errors, incidents <strong>and</strong> accidents in anaesthetic<br />
practice. Anaesth Intensive Care. 1993;21:506-519.<br />
15. Caplan RA, Posner KL, Cheney FW. Effect <strong>of</strong> outcome on physician judgments <strong>of</strong><br />
appropriateness <strong>of</strong> care. JAMA. 1991;265:1957-1960.<br />
16. Perrow C. Normal accidents : Living with High-Risk Technologies. With a New Afterword<br />
<strong>and</strong> a Postscript on the Y2K Problem. Princeton, NJ: Princeton University Press. 1999.<br />
55
17. Rasmussen J. Human error <strong>and</strong> the problem <strong>of</strong> causality in analysis <strong>of</strong> accidents. Philos<br />
Trans R Soc Lond B Biol Sci. 1990;327:449-460.<br />
18. Pope C, Mays N. Reaching the parts other methods cannot reach: an introduction to<br />
qualitative methods in health <strong>and</strong> health services research. BMJ. 1995;311:42-45.<br />
19. Giacomini MK, Cook DJ. Users' guides to the medical literature: XXIII. Qualitative<br />
research in health care A. Are the results <strong>of</strong> the study valid? Evidence-Based Medicine<br />
Working Group. JAMA. 2000;284:357-362.<br />
20. Giacomini MK, Cook DJ. Users' guides to the medical literature: XXIII. Qualitative<br />
research in health care B. What are the results <strong>and</strong> how do they help me care for my<br />
patients? Evidence-Based Medicine Working Group. JAMA. 2000;284:478-482.<br />
21. Van der Schaaf T. Near miss reporting in the Chemical Process Industry [Doctoral thesis].<br />
Eindhoven, The Netherl<strong>and</strong>s: Eindhoven University <strong>of</strong> Technology. 1992.<br />
22. Moray N. Error reduction as a systems problem. In: Bogner M, editor. Human error in<br />
medicine. Hillsdale, NJ: Lawrence Erlbaum Associates. 1994.<br />
23. Ishikawa K. What is total quality control? The Japanese way. Englewood Cliffs, NJ:<br />
Prentice Hall. 1985.<br />
24. Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Statistics.<br />
Available at: http://www.JCAHO.org. Accessed April 16, 2001.<br />
25. Kaplan HS, Battles JB, Van der Schaaf TW, Shea CE, Mercer SQ. Identification <strong>and</strong><br />
classification <strong>of</strong> the causes <strong>of</strong> events in transfusion medicine. Transfusion. 1998;38:1071-<br />
1081.<br />
26. Will hospitals come clean for JCAHO? Trustee. 1998;51:6-7.<br />
27. Leape LL, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems<br />
analysis <strong>of</strong> adverse drug events. ADE Prevention Study Group. JAMA. 1995;274:35-43.<br />
28. Bhasale A. The wrong diagnosis: identifying causes <strong>of</strong> potentially adverse events in<br />
general practice using incident monitoring. Fam Pract. 1998;15:308-318.<br />
29. Troidl H. Disasters <strong>of</strong> endoscopic surgery <strong>and</strong> how to avoid them: error analysis. World J<br />
Surg. 1999;23:846-855.<br />
30. Fern<strong>and</strong>es CM, Walker R, Price A, Marsden J, Haley L. Root cause analysis <strong>of</strong> laboratory<br />
delays to an emergency department. J Emerg Med. 1997;15:735-739.<br />
31. H<strong>of</strong>er TP, Kerr EA. What is an error? Eff Clin Pract. 2000;3:261-269.<br />
56
PART III. PATIENT SAFETY PRACTICES & TARGETS<br />
Section A. Adverse Drug Events (ADEs)<br />
Section B. Infection Control<br />
Section C. Surgery, Anesthesia, <strong>and</strong> Perioperative Medicine<br />
Section D. <strong>Safety</strong> <strong>Practices</strong> for Hospitalized or Institutionalized Elders<br />
Section E. General Clinical Topics<br />
Section F. Organization, Structure, <strong>and</strong> Culture<br />
Section G. Systems Issues <strong>and</strong> Human Factors<br />
Section H. Role <strong>of</strong> the <strong>Patient</strong><br />
57
Section A. Adverse Drug Events (ADEs)<br />
Chapter 6. Computerized Physician Order Entry (CPOE) with Clinical Decision Support<br />
Systems (CDSSs)<br />
Chapter 7. The Clinical Pharmacist’s Role in Preventing Adverse Drug Events<br />
Chapter 8. Computer Adverse Drug Event (ADE) Detection <strong>and</strong> Alerts<br />
Chapter 9. Protocols for High-Risk Drugs: Reducing Adverse Drug Events Related to<br />
Anticoagulants<br />
Chapter 10. Unit-Dose Drug Distribution Systems<br />
Chapter 11. Automated Medication Dispensing Devices<br />
58
Chapter 6. Computerized Physician Order Entry (CPOE) with Clinical<br />
Decision Support Systems (CDSSs)<br />
Rainu Kaushal, MD, MPH<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
Medication errors <strong>and</strong> adverse drug events (ADEs) are common, costly, <strong>and</strong> clinically<br />
important problems. 1-7 Two inpatient studies, one in adults <strong>and</strong> one in pediatrics, have found<br />
that about half <strong>of</strong> medication errors occur at the stage <strong>of</strong> drug ordering, 2, 7 although direct<br />
observation studies have indicated that many errors also occur at the administration stage. 8 The<br />
principal types <strong>of</strong> medication errors, apart from missing a dose, include incorrect medication<br />
dose, frequency, or route. 2 ADEs are injuries that result from the use <strong>of</strong> a drug. Systems-based<br />
analysis <strong>of</strong> medication errors <strong>and</strong> ADEs suggest that changes in the medication ordering system,<br />
including the introduction <strong>of</strong> computerized physician order entry (CPOE) with clinical decision<br />
support systems (CDSSs), may reduce medication-related errors. 9<br />
Despite growing evidence <strong>and</strong> public m<strong>and</strong>ates, implementation <strong>of</strong> CPOE has been<br />
limited. The Leapfrog Group, a consortium <strong>of</strong> companies from the Business Roundtable, has<br />
endorsed CPOE in hospitals as one <strong>of</strong> the 3 changes that would most improve patient safety in<br />
America (see also Chapter 55). 10 A Medicare Payment Advisory Commission report suggested<br />
instituting financial incentives for CPOE implementation. 11 US Senators Bob Graham (D-Fla.)<br />
<strong>and</strong> Olympia Snowe (R-Maine) recently introduced a bill, entitled the “Medication Errors<br />
Reduction Act <strong>of</strong> 2001,” to establish an informatics system grant program for hospitals <strong>and</strong><br />
skilled nursing facilities. 12 In addition, California recently enacted legislation stipulating that<br />
acute care hospitals implement information technology, such as CPOE to reduce medicationrelated<br />
errors. 13<br />
Practice Description<br />
CPOE refers to a variety <strong>of</strong> computer-based systems <strong>of</strong> ordering medications, which<br />
share the common features <strong>of</strong> automating the medication ordering process. Basic CPOE ensures<br />
st<strong>and</strong>ardized, legible, complete orders by only accepting typed orders in a st<strong>and</strong>ard <strong>and</strong> complete<br />
format. Almost all CPOE systems include or interface with CDSSs <strong>of</strong> varying sophistication.<br />
Basic clinical decision support may include suggestions or default values for drug doses, routes,<br />
<strong>and</strong> frequencies. More sophisticated CDSSs can perform drug allergy checks, drug-laboratory<br />
value checks, drug-drug interaction checks, in addition to providing reminders about corollary<br />
orders (eg, prompting the user to order glucose checks after ordering insulin) or drug guidelines<br />
to the physician at the time <strong>of</strong> drug ordering (see also Chapter 53).<br />
At times, CDSSs are implemented without CPOE. Isolated CDSSs can provide advice on<br />
drug selection, dosages, <strong>and</strong> duration. More refined CDSSs can incorporate patient-specific<br />
information (for example recommending appropriate anticoagulation regimens), or incorporate<br />
pathogen-specific information such as suggesting appropriate anti-infective regimens. After<br />
viewing such advice, the physician proceeds with a conventional h<strong>and</strong>written medication order.<br />
59
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
It is estimated that over 770,000 people are injured or die in hospitals from ADEs<br />
annually. 4, 5, 14 The few hospitals that have studied incidence rates <strong>of</strong> ADEs have documented<br />
rates ranging from 2 to 7 per 100 admissions. 2, 4, 15, 16 A precise national estimate is difficult to<br />
calculate due to the variety <strong>of</strong> criteria <strong>and</strong> definitions used by researchers. 17 One study <strong>of</strong><br />
preventable inpatient ADEs in adults demonstrated that 56% occurred at the stage <strong>of</strong> ordering,<br />
34% at administration, 6% at transcribing, <strong>and</strong> 4% at dispensing. 2 In this study, the drug class<br />
most commonly associated with preventable ADEs was analgesics, followed by sedatives <strong>and</strong><br />
antibiotics. Even fewer studies have been conducted in the outpatient setting. One recent crosssectional<br />
chart review <strong>and</strong> patient care survey found an ADE rate <strong>of</strong> 3% in adult primary care<br />
outpatients. 18<br />
Opportunities for Impact<br />
Clear data do not exist about the prevalence <strong>of</strong> CPOE with CDSSs or isolated CDSSs.<br />
One survey <strong>of</strong> 668 hospitals indicated that 15% had at least partially implemented CPOE. 19 A<br />
slightly more recent survey <strong>of</strong> pharmacy directors at 1050 acute care hospitals in the United<br />
States (51% response rate) reported that 13% <strong>of</strong> hospitals had an electronic medication orderentry<br />
system in place, while an additional 27% stated they were in the process <strong>of</strong> obtaining such<br />
a sysytem. 20<br />
Study Designs<br />
The 4 studies listed in Table 6.1 evaluated CPOE with CDSSs. 21-24 The first study, a<br />
r<strong>and</strong>omized controlled trial evaluating the utility <strong>of</strong> CPOE in improving prescribing for corollary<br />
orders, was conducted by the Regenstrief Institute for Health Care (affiliated with the Indiana<br />
University School <strong>of</strong> Medicine). 21 The remaining 3 studies evaluated the CPOE system at<br />
Brigham <strong>and</strong> Women’s Hospital (BWH). 22-24 Of note, both the Regenstrief <strong>and</strong> BWH systems<br />
are “home-grown” rather than “<strong>of</strong>f-the-shelf” commercial systems. The first BWH study was a<br />
cross-sectional analysis comparing an intervention period <strong>of</strong> CPOE with CDSSs for adult<br />
inpatients on medical, surgical, <strong>and</strong> intensive care wards with a historical period. 22 The other 2<br />
23, 24<br />
BWH studies were time series analyses <strong>of</strong> orders written for adult inpatients.<br />
Table 6.2 lists 4 studies 25-28 that evaluated isolated CDSSs, 2 <strong>of</strong> which represent<br />
systematic reviews. 25, 26 Hunt et al 25 updated an earlier systematic review, 29 <strong>and</strong> included 68<br />
studies that prospectively evaluated use <strong>of</strong> CDSSs in a clinical setting by a healthcare<br />
practitioner with a concurrent control. Importantly, Hunt et al included studies <strong>of</strong> outpatient<br />
settings in their review. Walton’s review addressed 15 articles that studied computerized advice<br />
on drug dosage for inpatients. 26 Evans et al, at Latter Day Saints Hospital, performed the other 2<br />
studies, 27, 28 again on a “home-grown” system. The first was a r<strong>and</strong>omized controlled trial <strong>of</strong><br />
empiric antibiotic selection using CDSSs. 27 The second study was a cross-sectional analysis<br />
comparing an intervention period <strong>of</strong> a computer-assisted management program for antiinfectives<br />
with a historical control period. 28 This second study 28 was likely excluded from the<br />
systematic review by Hunt et al 25 for methodologic reasons, as it did not have a concurrent<br />
control. The reasons for excluding the first study 27 remain unclear. Finally, it is important to<br />
emphasize again that <strong>of</strong> the primary studies that were included, 6 <strong>of</strong> 8 were performed at 3<br />
institutions with sophisticated “home-grown” systems.<br />
60
Study Outcomes<br />
Adverse drug events (ADEs), (injuries that result from the use <strong>of</strong> drugs) by definition<br />
constitute clinical outcomes (Level 1). ADEs that are associated with a medication error are<br />
considered preventable, while those not associated with a medication error (eg, known<br />
medication side effects) are considered non-preventable. An example <strong>of</strong> a preventable ADE is<br />
the development <strong>of</strong> rash after the administration <strong>of</strong> ampicillin to a known penicillin-allergic<br />
patient. In contrast, a non-preventable ADE would be development <strong>of</strong> an ampicillin-associated<br />
rash in a patient with no known drug allergies. Non-intercepted serious medication errors include<br />
non-intercepted potential ADEs <strong>and</strong> preventable ADEs (ie, medication errors that either have the<br />
potential or actually cause harm to a patient).<br />
Medication errors refer to errors in the processes <strong>of</strong> ordering, transcribing, dispensing,<br />
administering, or monitoring medications, irrespective <strong>of</strong> the outcome (ie, injury to the patient).<br />
One example is an order written for amoxicillin without a route <strong>of</strong> administration. Other<br />
medication errors have a greater potential for patient harm <strong>and</strong> so are <strong>of</strong>ten designated as<br />
“serious medication errors” or “potential ADEs” – eg, an order for amoxicillin in a patient with<br />
past anaphylaxis to penicillin.<br />
Potential ADEs may or may not be intercepted before reaching the patient. An example<br />
<strong>of</strong> an intercepted potential ADE would be an order written for an acetaminophen overdose that is<br />
intercepted <strong>and</strong> corrected by a nurse before reaching the patient. An example <strong>of</strong> a nonintercepted<br />
potential ADE would be an administered overdose <strong>of</strong> acetaminophen to a patient who<br />
did not suffer any sequelae.<br />
Medication errors include a mixture <strong>of</strong> serious medication errors with a significant<br />
potential for patient injury (Level 2) <strong>and</strong> other deviations from recommended practice that do not<br />
have a clear or established connection to adverse events (Level 3). Examples <strong>of</strong> the latter<br />
include failure to choose the optimal dosing schedule for a medication <strong>and</strong> many st<strong>and</strong>ards<br />
related to monitoring serum drug levels <strong>and</strong> routine electrolytes.<br />
Only 2 studies (from a single institution) evaluating CPOE with CDSSs included ADEs<br />
as a secondary outcome (Level 1), 22 <strong>and</strong> even these studies primary outcomes were serious<br />
medication errors (Level 2) <strong>and</strong> non-intercepted medication errors. 23 The other studies reported<br />
a variety <strong>of</strong> other errors <strong>of</strong>ten involving mixtures <strong>of</strong> Level 2 <strong>and</strong> Level 3 outcomes – eg,<br />
“prescribing practices” 24 <strong>and</strong> “corollary orders.” 21 Corollary orders refer to orders required to<br />
detect or ameliorate adverse reactions that may result from the trigger order - eg, checking the<br />
serum creatinine a minimum <strong>of</strong> 2 times per week in a patient receiving a nephrotoxic agent such<br />
as amphotericin. Many corollary orders capture Level 3 outcomes, as failure to prescribe these<br />
orders would in most cases have no clear impact on clinical outcomes – eg, failure to order daily<br />
tests for fecal occult blood in patients on heparin or screening audiometry for patients receiving<br />
vancomycin. 21<br />
The predominance <strong>of</strong> Level 2 <strong>and</strong> 3 outcomes in these studies is underst<strong>and</strong>able, given<br />
the much higher frequency <strong>of</strong> these outcomes compared to actual ADEs <strong>and</strong> the enormous costs<br />
<strong>of</strong> conducting these studies.<br />
Similarly, the studies evaluating CDSSs reported a mixture <strong>of</strong> outcomes from Levels 1-3.<br />
Hunt et al reviewed articles to determine changes in patient outcomes (Level 1) or physician<br />
performance (Levels 2 <strong>and</strong> 3, depending on the practice). 25 Walton et al evaluated a range <strong>of</strong><br />
outcomes (Levels 1-3), including reductions in “adverse reactions <strong>and</strong> unwanted effects” (Level<br />
1). 26 In one study, Evans et al determined rates <strong>of</strong> pathogen susceptibility to an antibiotic<br />
61
egimen 27 (Level 2); another study by the same authors evaluated adverse drug events caused by<br />
anti-infectives as a main outcome (Level 1). 28<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Of the 2 studies at BWH that addressed the impact <strong>of</strong> CPOE with CDSSs on medication<br />
errors <strong>and</strong> ADEs, the first demonstrated a 55% decrease in serious medication errors. 22 As a<br />
secondary outcome, this study found a 17% decrease in preventable ADEs, which was not<br />
statistically significant. The second study, a time series analysis, demonstrated marked<br />
reductions in all medication errors excluding missed dose errors <strong>and</strong> in non-intercepted serious<br />
medication errors. 23 The number <strong>of</strong> ADEs in this study was quite small – 25 in the baseline<br />
period <strong>and</strong> 18 in the third <strong>of</strong> the 3 periods with CPOE <strong>and</strong> CDSS. Correcting for the number <strong>of</strong><br />
opportunities for errors, the total number <strong>of</strong> ADEs/1000 patient days decreased from 14.7 to 9.6<br />
(p=0.09). For the sub-category <strong>of</strong> preventable ADEs, the reduction (from 5 to 2) achieved<br />
borderline statistical significance (p=0.05).<br />
Overhage et al <strong>and</strong> Teich et al studied more focused aspects <strong>of</strong> the medication system.<br />
Overhage et al 21 studied computerized reminders for corollary orders (eg, entering a laboratory<br />
order to check electrolytes when ordering potassium for a patient) <strong>and</strong> showed a greater than<br />
100% improvement in the rate <strong>of</strong> corollary orders (p
Potential for Harm<br />
Faulty decision support data, for example an incorrect default dosing suggestion, can lead<br />
to inappropriate ordering choices by physicians. The BWH time series analysis demonstrated an<br />
initial rise in intercepted potential ADEs due to the structure <strong>of</strong> the ordering screen for potassium<br />
chloride. 23 This structural error was identified <strong>and</strong> easily rectified, but underscores the<br />
importance <strong>of</strong> close scrutiny <strong>of</strong> all aspects <strong>of</strong> CPOE screens, both initially <strong>and</strong> on an ongoing<br />
basis. 23<br />
Also, analogously to writing an order in the wrong patient chart in a conventional system,<br />
a physician can electronically write an order in the wrong patient’s record - eg, after walking<br />
away from the terminal, opening the wrong record from a personalized rounding list. In<br />
addition, it is critical that the trigger level for warnings appropriately balances alarm sensitivity<br />
<strong>and</strong> specificity. These systems must have thresholds set so that physicians receive warnings in<br />
situations with a potential for significant harm, without being overwhelmed by “false alarms.”<br />
Another potential risk is hardware outage or s<strong>of</strong>tware instability. For example, the reliability<br />
that is needed with CPOE is much higher than that required for systems that simply report<br />
laboratory results.<br />
Costs <strong>and</strong> Implementation<br />
Six <strong>of</strong> the studies described in this review evaluated “home-grown” rather than “<strong>of</strong>f-theshelf”<br />
systems. The present costs for purchasing commercial systems are substantially more<br />
than the previous costs <strong>of</strong> developing such systems. For BWH, the cost <strong>of</strong> developing <strong>and</strong><br />
implementing CPOE in 1992 was estimated to be $1.9 million, with maintenance costs <strong>of</strong><br />
$500,000 per year. In comparison, the cost <strong>of</strong> purchasing <strong>and</strong> implementing large commercial<br />
systems varies substantially, but may be on the order <strong>of</strong> tens <strong>of</strong> millions <strong>of</strong> dollars. Several<br />
studies demonstrate that only minimal resources are needed to introduce <strong>and</strong>/or maintain<br />
21, 24, 30<br />
decision support programs into existing order entry programs.<br />
Relatively few data are available regarding the financial benefits <strong>of</strong> CPOE, although they<br />
extend well beyond medication-related events. The net savings <strong>of</strong> the BWH system are<br />
estimated at $5-10 million per year. 31 It is estimated that the costs to BWH for preventable<br />
ADEs are $2.8 million annually. 32 Evans et al reported a $100,000 per year cost avoidance with<br />
a computer-assisted antibiotic dosing program largely attributable to decreased antibiotic use <strong>and</strong><br />
avoided ADEs. 33<br />
Importantly, healthcare systems must garner both financial <strong>and</strong> organizational support<br />
before introducing CPOE with CDSSs. CPOE requires a very large up-front investment with<br />
more remote, albeit substantial returns. In addition, CPOE impacts clinicians <strong>and</strong> workflow<br />
substantially. Its complexity requires close integration with multiple systems, such as the<br />
laboratory <strong>and</strong> pharmacy systems. Failure to attend to the impact <strong>of</strong> such a large-scale effort on<br />
organizational culture <strong>and</strong> dynamics may result in implementation failure. 34 Therefore, it is<br />
essential to have organizational support <strong>and</strong> integration for its successful implementation <strong>and</strong><br />
use.<br />
Comment<br />
The literature supports CPOE’s beneficial effect in reducing the frequency <strong>of</strong> a range <strong>of</strong><br />
medication errors, including serious errors with the potential for harm. Fewer data are available<br />
regarding the impact <strong>of</strong> CPOE on ADEs, with no study showing a significant decrease in actual<br />
patient harm. Similarly, isolated CDSSs appear to prevent a range <strong>of</strong> medication errors, but with<br />
63
few data describing reductions in ADEs or improvements in other clinical outcomes. Finally, the<br />
studied CDSSs address focused medication use (for example, antibiotic dosing) rather than more<br />
general aspects <strong>of</strong> medication use.<br />
Further research should be conducted to compare the various types <strong>of</strong> systems <strong>and</strong> to<br />
compare “home-grown” with commercially available systems. Such comparisons are<br />
particularly important since the institutions that have published CPOE outcomes have generally<br />
been those with strong institutional commitments to their systems. Whether less committed<br />
institutions purchasing “<strong>of</strong>f the shelf” systems will see benefits comparable to those enjoyed by<br />
“pioneers” with home-grown systems remains to be determined. Studying the benefits <strong>of</strong> such<br />
complex systems requires rigorous methodology <strong>and</strong> sufficient size to provide the power to<br />
study ADEs. Further research also needs to address optimal ways for institutions to acquire <strong>and</strong><br />
implement computerized ordering systems.<br />
64
Table 6.1. Studies <strong>of</strong> computerized physician order entry (CPOE) with clinical decision<br />
support systems (CDSSs) *<br />
Study Study Design Study Outcomes Results<br />
Overhage, 1997. 21 Impact <strong>of</strong><br />
faculty <strong>and</strong> physician<br />
reminders (using CPOE) on<br />
corollary orders for adult<br />
inpatients in a general medical<br />
ward at a public teaching<br />
hospital affiliated with the<br />
Indiana University School <strong>of</strong><br />
Medicine<br />
Bates, 1998. 22 CPOE with<br />
CDSSs for adult inpatients on<br />
medical, surgical, <strong>and</strong> intensive<br />
care wards at BWH, a tertiary<br />
care center affiliated with<br />
Harvard University<br />
Bates, 1999. 23 CPOE with<br />
CDSSs for adult inpatients in 3<br />
medical units at BWH<br />
Teich, 2000. 24 CPOE with<br />
CDSSs for all adult inpatients at<br />
BWH<br />
Level 1<br />
(RCT with<br />
physicians<br />
r<strong>and</strong>omized<br />
to receive<br />
reminders or<br />
not)<br />
Levels 2 & 3<br />
(two study<br />
designs)<br />
Level 3<br />
(retrospective<br />
time series)<br />
Level 3<br />
(retrospective<br />
before-after<br />
analysis)<br />
65<br />
Levels 2 & 3<br />
(errors <strong>of</strong> omission<br />
in corollary orders)<br />
Level 1 (ADE<br />
rates) <strong>and</strong> Level 2<br />
(serious<br />
medication errors)<br />
Level 1 (ADEs)<br />
<strong>and</strong> Level 2 (main<br />
outcome measure<br />
was medication<br />
errors)<br />
Levels 2 & 3<br />
(changes in 5<br />
prescribing<br />
practices)<br />
25% improvement in<br />
ordering <strong>of</strong> corollary<br />
medications by faculty<br />
<strong>and</strong> residents<br />
(p
Table 6.2. Studies <strong>of</strong> clinical decision support systems (CDSSs) *<br />
Study Setting Study Design Study Outcomes Results<br />
Hunt, 1998. 25 Use <strong>of</strong><br />
CDSSs by healthcare<br />
practitioners in<br />
multiple inpatient <strong>and</strong><br />
outpatient settings<br />
Walton, 2001. 26 Use <strong>of</strong><br />
CDSSs for drug dosage<br />
advice by healthcare<br />
practitioners for 1229<br />
patients in multiple<br />
inpatient settings<br />
Evans, 1994. 27 Use <strong>of</strong> a<br />
computerized antibiotic<br />
selection consultant for<br />
451 inpatients at Salt<br />
Lake City’s LDS<br />
Hospital, a 520-bed<br />
community teaching<br />
hospital <strong>and</strong> tertiary<br />
referral center<br />
Evans, 1998. 28<br />
Computer-based antiinfective<br />
management<br />
program for 1136<br />
intensive care unit<br />
patients from a 12-bed<br />
ICU at LDS Hospital<br />
Level 1A<br />
(systematic<br />
review <strong>of</strong> RCTs)<br />
Levels 1A-3A<br />
(systematic<br />
review <strong>of</strong> RCTs,<br />
interrupted time<br />
series analyses,<br />
<strong>and</strong> controlled<br />
before-after<br />
studies)<br />
Level 1 (RCT<br />
with crossover<br />
design)<br />
Level 2<br />
(prospective<br />
before-after<br />
analysis)<br />
Levels 1 & 2 (a<br />
variety <strong>of</strong> measures<br />
related to patient<br />
outcomes <strong>and</strong><br />
physician practice, not<br />
just ADEs <strong>and</strong><br />
processes <strong>of</strong> care<br />
related to medication<br />
use.<br />
Level 1 (one main<br />
outcome measure was<br />
adverse drug reactions<br />
Level 2 (one <strong>of</strong> 5<br />
primary outcomes was<br />
pathogen<br />
susceptibility to<br />
prescribed antibiotic<br />
regimens<br />
Level 2 (one primary<br />
outcome was ADEs<br />
due to anti-infective<br />
agents<br />
66<br />
6 <strong>of</strong> 14 studies showed<br />
improvement in patient<br />
outcomes<br />
43 <strong>of</strong> 65 studies<br />
showed improvement<br />
in physician<br />
performance<br />
Absolute risk reduction<br />
with CDSSs: 6% (95%<br />
CI: 0-12%)<br />
17% greater pathogen<br />
susceptibility to an<br />
antibiotic regimen<br />
suggested by computer<br />
consultant versus<br />
physicians (p
References<br />
1. Barker KN, Mikeal RL, Pearson RE, Illig NA, Morse ML. Medication errors in nursing<br />
homes <strong>and</strong> small hospitals. Am J Hosp Pharm. 1982; 39: 987-91.<br />
2. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong><br />
adverse drug events <strong>and</strong> potential adverse drug events: Implications for prevention.<br />
JAMA. 1995; 274: 29-34.<br />
3. Dean BS, Allan EL, Barber ND, Barker KN. Comparison <strong>of</strong> medication errors in an<br />
American <strong>and</strong> British hospital. Am J Health Syst Pharm. 1995; 52: 2543-49.<br />
4. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in<br />
hospitalized patients. Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA.<br />
1997; 277: 301-6.<br />
5. Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable<br />
adverse drug events in hospitalized patients: a comparative study <strong>of</strong> intensive care <strong>and</strong><br />
general care units. Crit Care Med. 1997; 25: 1289-97.<br />
6. Lesar TS, Lomaestro BM, Pohl H. Medication-prescribing errors in a teaching hospital.<br />
A 9-year experience. Arch Intern Med. 1997; 157: 1569-76.<br />
7. Kaushal R, Bates DW, L<strong>and</strong>rigan C, McKenna J, Clapp MD, Federico F, et al.<br />
Medication Errors <strong>and</strong> Adverse Drug Events in Pediatric Inpatients. JAMA. 2001; 285:<br />
2114-20.<br />
8. Allan EL, Barker KN. Fundamentals <strong>of</strong> medication error research. Am J Hosp Pharm.<br />
1990; 47: 555-71.<br />
9. Leape LL, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems<br />
analysis <strong>of</strong> adverse drug events. ADE Prevention Study Group. JAMA. 1995; 274: 35-43.<br />
10. Milstein A, Galvin RS, Delbanco SF, Salber P, Buck Jr CR. Improving the safety <strong>of</strong><br />
health care: the leapfrog initiative. Eff Clin Pract. 2000; 3: 313-6.<br />
11. The Medicare Payment Advisory Commission. Report to the Congress. Selected<br />
Medicare Issues. Available at: http://www.medpac.gov/. Accessed May 21, 2001.<br />
12. United States Senate. Press Release. Available at:<br />
http://www.senate.gov/~graham/pr050301.html. Accessed June 12, 2001.<br />
13. California Senate Bill No. 1875. Chapter 816, Statutes <strong>of</strong> 2000. (Sept. 28, 2000)<br />
14. Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal AR, Leape LL. The incident<br />
reporting system does not detect adverse drug events: a problem for quality<br />
improvement. Jt Comm J Qual Improv. 1995; 21: 541-8.<br />
67
15. Bates DW, Boyle DL, V<strong>and</strong>er Vliet MB, Schneider J, Leape L. Relationship between<br />
medication errors <strong>and</strong> adverse drug events. J Gen Intern Med. 1995; 10: 199-205.<br />
16. Jha AK, Kuperman GJ, Teich JM, Leape L, Shea B, Rittenberg E, et al. Identifying<br />
adverse drug events: development <strong>of</strong> a computer-based monitor <strong>and</strong> comparison with<br />
chart review <strong>and</strong> stimulated voluntary report. J Am Med Inform Assoc. 1998; 5: 305-14.<br />
17. Adverse drug events: the magnitude <strong>of</strong> health risk is uncertain because <strong>of</strong> limited<br />
incidence data. Washington, D.C.: General Accounting Office; 2000.<br />
18. G<strong>and</strong>hi TK, Burstin HR, Cook EF, Puopolo AL, Haas JS, Brennan TA, et al. Drug<br />
Complications in Outpatients. J Gen Intern Med. 2000; 15: 149-54.<br />
19. Ash JS, Gorman PN, Hersh WR. Physician order entry in U.S. hospitals. Proceedings<br />
AMIA Annual Symposium. 1998: 235-39.<br />
20. Ringold DJ, Santell JP, Schneider PJ. ASHP national survey <strong>of</strong> pharmacy practice in<br />
acute care settings: dispensing <strong>and</strong> administration--1999. Am J Health Syst Pharm. 2000;<br />
57: 1759-75.<br />
21. Overhage JM, Tierney WM, Zhou XH, McDonald CJ. A r<strong>and</strong>omized trial <strong>of</strong> "corollary<br />
orders" to prevent errors <strong>of</strong> omission. J Am Med Inform Assoc. 1997; 4: 364-75.<br />
22. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA, Teich JM, et al. Effect <strong>of</strong><br />
computerized physician order entry <strong>and</strong> a team intervention on prevention <strong>of</strong> serious<br />
medication errors . JAMA. 1998; 280: 1311-6.<br />
23. Bates DW, Teich JM, Lee J, Seger D, Kuperman GJ, Ma'Luf N, et al. The impact <strong>of</strong><br />
computerized physician order entry on medication error prevention. J Am Med Inform<br />
Assoc. 1999; 6: 313-21.<br />
24. Teich JM, Merchia PR, Schmiz JL, Kuperman GJ, Spurr CD, Bates DW. Effects <strong>of</strong><br />
computerized physician order entry on prescribing practices. Arch Intern Med. 2000;<br />
160: 2741-7.<br />
25. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects <strong>of</strong> computer-based clinical decision<br />
support systems on physician performance <strong>and</strong> patient outcomes: a systematic review.<br />
JAMA. 1998; 280: 1339-46.<br />
26. Walton RT, Harvey E, Dovey S, Freemantle N. Computerised advice on drug dosage to<br />
improve prescribing practice (Cochrane Review). In: The Cochrane Library, Issue 2,<br />
2001. Oxford: Update S<strong>of</strong>tware.<br />
27. Evans RS, Classen DC, Pestotnik SL, Lundsgaarde HP, Burke JP. Improving empiric<br />
antibiotic selection using computer decision support. Arch Intern Med. 1994; 154: 878-<br />
84.<br />
68
28. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF, Jr., et al. A<br />
computer-assisted management program for antibiotics <strong>and</strong> other antiinfective agents. N<br />
Engl J Med. 1998; 338: 232-8.<br />
29. Johnston ME, Langton KB, Haynes RB, Mathieu A. Effects <strong>of</strong> computer-based clinical<br />
decision support systems on clinician performance <strong>and</strong> patient outcome. A critical<br />
appraisal <strong>of</strong> research . Ann Intern Med. 1994; 120: 135-42.<br />
30. Shojania KG, Yokoe D, Platt R, Fiskio J, Ma'luf N, Bates DW. Reducing vancomycin<br />
use utilizing a computer guideline: results <strong>of</strong> a r<strong>and</strong>omized controlled trial. J Am Med<br />
Inform Assoc. 1998; 5: 554-62.<br />
31. Teich JM, Glaser JP, Beckley RF, Aranow M, Bates DW, Kuperman GJ, et al. Toward<br />
cost-effective, quality care; the Brigham Integrated Computing System. Proc. 2nd<br />
Nicholas E. Davies CPR Recognition Symposium. Washington, D.C.: Computerized<br />
<strong>Patient</strong> Record Institute. 1996;3-34.; 1996: 3-34.<br />
32. Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA, et al. The costs <strong>of</strong><br />
adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study<br />
Group. JAMA. 1997; 277: 307-11.<br />
33. Evans RS, Pestonik SL, Classen DC, Burke JP. Evaluation <strong>of</strong> a computer-assisted<br />
antibiotic-dose monitor. Ann Pharmacother. 1999; 33: 1026-31.<br />
34. Massaro TA. Introducing physician order entry at a major academic medical center: I.<br />
Impact on organizational culture <strong>and</strong> behavior. Acad Med. 1993; 68: 20-5.<br />
69
Chapter 7. The Clinical Pharmacist’s Role in Preventing Adverse Drug<br />
Events<br />
Rainu Kaushal, MD, MPH<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
A large literature documents the multiple roles clinical pharmacists can play in a variety<br />
<strong>of</strong> health care settings. 1-8 Much <strong>of</strong> this literature focuses on measures <strong>of</strong> impact not directly<br />
relevant to this Report – eg, economic benefits, 4,8 patient compliance, 6 <strong>and</strong> drug monitoring. 2,3<br />
More recently, systems-based analyses <strong>of</strong> medication errors <strong>and</strong> adverse drug events (ADEs)<br />
have drawn attention to the impact clinical pharmacists can exert on the quality <strong>and</strong> safety <strong>of</strong><br />
medication use. 9-12 In this chapter, we review the evidence supporting the premise that direct<br />
participation <strong>of</strong> pharmacists’ in clinical care reduces medication errors <strong>and</strong> ADEs in hospitalized<br />
<strong>and</strong> ambulatory patients.<br />
Practice Description<br />
Clinical pharmacists may participate in all stages <strong>of</strong> the medication use process,<br />
including drug ordering, transcribing, dispensing, administering, <strong>and</strong> monitoring. The specific<br />
activities <strong>of</strong> clinical pharmacists vary substantially in the studies reviewed in this chapter. In the<br />
hospital setting, one study evaluated the role <strong>of</strong> a senior pharmacist participating fully in<br />
intensive care unit rounds <strong>and</strong> available throughout the day in person or by page for questions. 13<br />
Another study evaluated a ward pharmacy service that examined order sheets for new therapies<br />
<strong>and</strong> carried out checks that were formerly performed in the pharmacy. 14<br />
Pharmacists may also play a role at the time <strong>of</strong> discharge. One study reported the impact<br />
<strong>of</strong> clinical pharmacists’ consultation for geriatric patients at the time <strong>of</strong> discharge, 15 with<br />
pharmacists serving as consultants to physicians <strong>and</strong> reinforcing patients’ knowledge <strong>of</strong> their<br />
medication regimen. The roles <strong>of</strong> clinical pharmacists are similarly diverse in studies <strong>of</strong><br />
ambulatory settings. Here they include the provision <strong>of</strong> consultative services, 16-18 patient<br />
education, 16-18 therapeutic drug monitoring, 3 <strong>and</strong> even follow-up telephone calls to patients. 3,16-18<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
It is estimated that over 770,000 people are injured or die in hospitals from adverse drug<br />
events (ADEs) annually. 19-21 The few hospitals that have studied incidence rates <strong>of</strong> ADEs have<br />
documented rates ranging from 2 to 7 per 100 admissions. 11,19,22,23 A precise national estimate is<br />
difficult to calculate due to the variety <strong>of</strong> criteria <strong>and</strong> definitions used by researchers. 24 One<br />
study <strong>of</strong> preventable inpatient ADEs in adults demonstrated that 56% occurred at the stage <strong>of</strong><br />
ordering, 34% at administration, 6% at transcribing, <strong>and</strong> 4% at dispensing. 22 In this study, the<br />
drug class most commonly associated with preventable ADEs was analgesics, followed by<br />
sedatives <strong>and</strong> antibiotics. Even fewer studies have been conducted in the outpatient setting. One<br />
recent cross-sectional chart review <strong>and</strong> patient care survey found an ADE rate <strong>of</strong> 3% in adult<br />
primary care outpatients. 25<br />
Opportunities for Impact<br />
Although many hospital pharmacy departments <strong>of</strong>fer clinical pharmacy consultation<br />
services, 26 the degree to which these services include rounding with clinicians 13 is unclear.<br />
71
Hospital pharmacies provide support to a variable degree in different ambulatory settings<br />
(clinics, nursing homes, adult daycare), 26 but the precise nature <strong>of</strong> clinical pharmacists’ activities<br />
in these settings is not uniformly characterized.<br />
Study Designs<br />
We identified 3 systematic reviews <strong>of</strong> clinical pharmacists in the outpatient setting. 5,17,18<br />
We included only the most recent review, 18 as it followed the most rigorous methodology <strong>and</strong><br />
included the bibliographies <strong>of</strong> the previous reviews. 5,17 We identified only one study 16 <strong>of</strong> the<br />
impact <strong>of</strong> clinical pharmacists on patient outcomes in the ambulatory setting that had not been<br />
included in this systematic review. 18 This study, a r<strong>and</strong>omized controlled trial evaluating clinical<br />
pharmacists’ participation in the management <strong>of</strong> outpatients with congestive heart failure, 16 was<br />
therefore also included.<br />
For studies <strong>of</strong> clinical pharmacists in the hospital setting, an older review 2 did not meet<br />
the characteristics <strong>of</strong> a systematic review, but was a very thorough summary <strong>of</strong> the relevant<br />
literature. It included 8 studies evaluating pharmacists’ roles in detecting <strong>and</strong> reporting ADEs in<br />
the hospital setting. Preliminary review <strong>of</strong> these studies revealed that the measured outcomes<br />
were Level 3 (detection <strong>of</strong> ADEs as an end in itself). Consequently, we did not include them.<br />
One older retrospective before-after analysis (Level 3) 14 did meet our inclusion criteria, as did 2<br />
more recent studies <strong>of</strong> clinical pharmacists’ roles in the inpatient setting: a prospective,<br />
controlled before-after study (Level 2) 13 <strong>and</strong> a r<strong>and</strong>omized controlled trial (Level 1). 15<br />
We included an additional meta-analysis focusing on therapeutic drug monitoring by<br />
clinical pharmacists in the hospital <strong>and</strong> ambulatory settings. 3 The studies included in this metaanalysis<br />
consisted predominantly <strong>of</strong> controlled observational studies (Level 3) <strong>and</strong> nonr<strong>and</strong>omized<br />
clinical trials (Level 2), but one r<strong>and</strong>omized controlled trial was also included (Level<br />
1-3A).<br />
Table 7.1 lists the studies reviewed in this chapter <strong>and</strong> briefly describes their salient<br />
features. All <strong>of</strong> the listed studies involved adult patients, as the single pediatric study 9 identified<br />
by our literature search had no control group (Level 4 design) <strong>and</strong> was therefore excluded.<br />
Study Outcomes<br />
Level 1 outcomes reported in the included studies consisted <strong>of</strong> ADEs 13 <strong>and</strong> clinical<br />
events related to heart failure, including mortality. 16 One study used telephone interviews to<br />
solicit patient self-reports <strong>of</strong> adverse drug reactions. 13,27 This study was excluded since the<br />
systematic review <strong>of</strong> clinical pharmacists’ roles in ambulatory settings 18 incorporated its findings<br />
in several <strong>of</strong> its analyses.<br />
The distinction between Level 2 <strong>and</strong> 3 outcomes can be somewhat ambiguous in studies<br />
<strong>of</strong> prescribing practice, as the exact point at which choices <strong>of</strong> therapeutic agents or dosing<br />
patterns become not just sub-optimal but actually represent “errors” is difficult to define<br />
precisely. Even for objective outcomes, such as serum drug concentrations, the connection to<br />
patient outcomes is weak in some cases (eg, monitoring vancomycin levels 28,29 ), <strong>and</strong> therefore<br />
more appropriately designated as Level 3 rather than Level 2. Acknowledging the subjectivity <strong>of</strong><br />
this judgment in some cases, we included studies that contained a mixture <strong>of</strong> Level 2 <strong>and</strong> 3<br />
outcomes, including “prescribing problems,” (which included inappropriate choice <strong>of</strong> therapy,<br />
dosage errors, frequency errors, drug-drug interactions, therapeutic duplications, <strong>and</strong> allergies 15 ),<br />
<strong>and</strong> serum drug concentrations for a broad range <strong>of</strong> medications. 3<br />
72
Evidence for Effectiveness <strong>of</strong> the Practice<br />
In the inpatient setting, Leape’s study 13 demonstrated a statistically significant 66%<br />
decrease in preventable ADEs due to medication ordering. The study <strong>of</strong> geriatric patients at the<br />
time <strong>of</strong> discharge demonstrated clinically <strong>and</strong> statistically significant decreases in medication<br />
errors. 15 The meta-analysis <strong>of</strong> the effect <strong>of</strong> clinical pharmacokinetics services on maintaining<br />
acceptable drug ranges indicated only modest effect sizes for the outcomes measured, <strong>and</strong> only 2<br />
<strong>of</strong> the main results achieved statistical significance. 3<br />
The comprehensive review <strong>of</strong> clinical pharmacist services in ambulatory settings reported<br />
positive impacts for patients with hypertension, hypercholesterolemia, chronic heart failure, <strong>and</strong><br />
diabetes. 18 However, the authors identify important limitations: these studies are not easily<br />
generalizable, only 2 studies compared pharmacist services with other health pr<strong>of</strong>essional<br />
services, <strong>and</strong> both studies had important biases. Consequently, they emphasized the need for<br />
more rigorous research to document the effects <strong>of</strong> outpatient pharmacist interventions. 18 The<br />
additional study <strong>of</strong> outpatients demonstrated significant decreases in mortality <strong>and</strong> heart failure<br />
events, 16 but these results may reflect closer follow-up <strong>and</strong> monitoring (including telemetry) for<br />
the intervention group or the higher doses <strong>of</strong> angiotensin-converting enzyme (ACE) inhibitors<br />
the patients received. Generalizing this benefit to other conditions is difficult since most<br />
conditions do not have a single medication-related process <strong>of</strong> care that delivers the marked<br />
clinical benefits as do ACE inhibitors in the treatment <strong>of</strong> congestive heart failure. 30<br />
Potential for Harm<br />
Introducing clinical pharmacists might potentially disrupt routine patient care activities.<br />
However, the 2 studies that assessed physician reactions to clinical pharmacists 13,27 found<br />
excellent receptivity <strong>and</strong> subsequent changes in prescribing behavior.<br />
Costs <strong>and</strong> Implementation<br />
Two studies examined resource utilization <strong>and</strong> cost savings in the inpatient setting. The<br />
intensive care unit study indicated that there could be potential savings <strong>of</strong> $270,000 per year for<br />
this hospital if the intervention involved re-allocation <strong>of</strong> existing pharmacists’ time <strong>and</strong><br />
resources. 13 McMullin et al studied all interventions made by 6 hospital pharmacists over a onemonth<br />
period at a large university hospital <strong>and</strong> estimated annual cost savings <strong>of</strong> $394,000. 31<br />
A systematic review <strong>of</strong> outpatient pharmacists indicated that pharmacist interventions<br />
could lead to increased scheduled service utilization <strong>and</strong> decreased non-scheduled service<br />
utilization, specialty visits, <strong>and</strong> numbers <strong>and</strong> costs <strong>of</strong> drugs. 18<br />
73
Comment<br />
At present, one study provides strong evidence for the benefit <strong>of</strong> clinical pharmacists in<br />
reducing ADEs in hospitalized intensive care unit patients. 13 One additional study provides<br />
modest support for the impact <strong>of</strong> ward-based clinical pharmacists on the safety <strong>and</strong> quality <strong>of</strong><br />
inpatient medication use. 14 The evidence in the outpatient setting is less substantial, <strong>and</strong> not yet<br />
convincing. Given the other well-documented benefits <strong>of</strong> clinical pharmacists <strong>and</strong> the promising<br />
results in the inpatient setting, more focused research documenting the impact <strong>of</strong> clinical<br />
pharmacist interventions on medication errors <strong>and</strong> ADEs is warranted.<br />
74
Table 7.1. Studies <strong>of</strong> clinical pharmacists’ impact on ADEs <strong>and</strong> medication errors*<br />
Study Study Design Study Outcomes Results<br />
Beney, 2000. 18 Systematic<br />
review <strong>of</strong> the roles <strong>and</strong> impacts<br />
<strong>of</strong> pharmacists in ambulatory<br />
settings; reviewed studies<br />
included 16,000 outpatients<br />
<strong>and</strong> 40 pharmacists<br />
Gattis, 1999. 16 181 patients<br />
with heart failure due to left<br />
ventricular dysfunction<br />
followed in a general<br />
cardiology clinic<br />
Leape, 1999. 13 <strong>Medical</strong> <strong>and</strong><br />
cardiac intensive care unit<br />
patients at Massachusetts<br />
General Hospital, a tertiary<br />
care hospital in Boston<br />
Leach, 1981. 14 315 patients at<br />
Queen Elizabeth Hospital in<br />
Birmingham, Engl<strong>and</strong><br />
Lipton, 1992. 15 236 geriatric<br />
patients discharged from the<br />
hospital on three or more<br />
medications<br />
Ried, 1989. 3 Pooled patient<br />
population not reported, but<br />
review <strong>of</strong> articles indicates<br />
predominance <strong>of</strong> studies <strong>of</strong><br />
(mostly adult) hospitalized<br />
patients<br />
Level 1A<br />
(systematic<br />
review)<br />
Level 1<br />
(RCT)<br />
Level 2<br />
(prospective<br />
before-after study<br />
with concurrent<br />
control)<br />
Level 3<br />
(retrospective<br />
before-after<br />
analysis)<br />
Levels 1-3<br />
(variety <strong>of</strong> patient<br />
outcomes, surrogate<br />
outcomes, impacts on<br />
physician prescribing<br />
practices <strong>and</strong> measures <strong>of</strong><br />
resource use)<br />
Level 1<br />
(mortality <strong>and</strong> other<br />
clinical outcomes related<br />
to heart failure)<br />
75<br />
Improvement in outcomes for<br />
patients with hypertension,<br />
hypercholesterolemia,<br />
chronic heart failure, <strong>and</strong><br />
diabetes<br />
16 versus 4 deaths or other<br />
heart failure events (p
References<br />
1. Dyer CC, Oles KS, Davis SW. The role <strong>of</strong> the pharmacist in a geriatric nursing home: a<br />
literature review. Drug Intell Clin Pharm. 1984;18:428-433.<br />
2. Hatoum HT, Catizone C, Hutchinson RA, Purohit A. An eleven-year review <strong>of</strong> the<br />
pharmacy literature: documentation <strong>of</strong> the value <strong>and</strong> acceptance <strong>of</strong> clinical pharmacy. Drug<br />
Intell Clin Pharm. 1986;20:33-48.<br />
3. Ried LD, McKenna DA, Horn JR. Meta-analysis <strong>of</strong> research on the effect <strong>of</strong> clinical<br />
pharmacokinetics services on therapeutic drug monitoring. Am J Hosp Pharm.<br />
1989;46:945-951.<br />
4. Willett MS, Bertch KE, Rich DS, Ereshefsky L. Prospectus on the economic value <strong>of</strong><br />
clinical pharmacy services. A position statement <strong>of</strong> the American College <strong>of</strong> Clinical<br />
Pharmacy. Pharmacotherapy. 1989;9:45-56.<br />
5. Carter BL, Helling DK. Ambulatory care pharmacy services: the incomplete agenda. Ann<br />
Pharmacother. 1992;26:701-708.<br />
6. Tett SE, Higgins GM, Armour CL. Impact <strong>of</strong> pharmacist interventions on medication<br />
management by the elderly: a review <strong>of</strong> the literature. Ann Pharmacother. 1993;27:80-86.<br />
7. Jenkins MH, Bond CA. The impact <strong>of</strong> clinical pharmacists on psychiatric patients.<br />
Pharmacotherapy. 1996;16:708-714.<br />
8. Schumock GT, Meek PD, Ploetz PA, Vermeulen LC. Economic evaluations <strong>of</strong> clinical<br />
pharmacy services–1988-1995. The <strong>Public</strong>ations Committee <strong>of</strong> the American College <strong>of</strong><br />
Clinical Pharmacy. Pharmacotherapy. 1996;16:1188-1208.<br />
9. Folli HL, Poole RL, Benitz WE, Russo JC. Medication error prevention by clinical<br />
pharmacists in two children's hospitals. Pediatrics. 1987;79:718-722.<br />
10. Lesar TS, Bricel<strong>and</strong> LL, Delcoure K, Parmalee JC, Masta-Gornic V, Pohl H. Medication<br />
prescribing errors in a teaching hospital. JAMA. 1990;263:2329-2334.<br />
11. Bates DW, Boyle DL, V<strong>and</strong>er Vliet MB, Schneider J, Leape L. Relationship between<br />
medication errors <strong>and</strong> adverse drug events. J Gen Intern Med. 1995;10:199-205.<br />
12. Kaushal R, Bates DW, L<strong>and</strong>rigan C, McKenna J, Clapp MD, Federico F, et al. medication<br />
errors <strong>and</strong> adverse drug events in pediatric inpatients. JAMA. 2001;285:2114-2120.<br />
13. Leape LL, Cullen DJ, Clapp MD, Burdick E, Demonaco HJ, Erickson JI, et al. Pharmacist<br />
participation on physician rounds <strong>and</strong> adverse drug events in the intensive care unit. JAMA.<br />
1999;282:267-270.<br />
14. Leach RH, Feetam C, Butler D. An evaluation <strong>of</strong> a ward pharmacy service. J Clin Hosp<br />
Pharm. 1981;6:173-182.<br />
15. Lipton HL, Bero LA, Bird JA, McPhee SJ. The impact <strong>of</strong> clinical pharmacists'<br />
consultations on physicians' geriatric drug prescribing. A r<strong>and</strong>omized controlled trial. Med<br />
Care. 1992;30:646-658.<br />
16. Gattis WA, Hasselblad V, Whellan DJ, O'Connor CM. Reduction in heart failure events by<br />
the addition <strong>of</strong> a clinical pharmacist to the heart failure management team: results <strong>of</strong> the<br />
Pharmacist in Heart Failure Assessment Recommendation <strong>and</strong> Monitoring (PHARM)<br />
Study. Arch Intern Med. 1999;159:1939-1945.<br />
17. Hatoum HT, Akhras K. 1993 Bibliography: a 32-year literature review on the value <strong>and</strong><br />
acceptance <strong>of</strong> ambulatory care provided by pharmacists. Ann Pharmacother.<br />
1993;27:1106-1119.<br />
76
18. Beney J, Bero LA, Bond C. Exp<strong>and</strong>ing the roles <strong>of</strong> outpatient pharmacists: effects on<br />
health services utilisation, costs, <strong>and</strong> patient outcomes. In: The Cochrane Library, Issue 2,<br />
2000. Oxford: Update S<strong>of</strong>tware.<br />
19. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in<br />
hospitalized patients. Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA.<br />
1997;277:301-306.<br />
20. Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal AR, Leape LL. The incident<br />
reporting system does not detect adverse drug events: a problem for quality improvement.<br />
Jt Comm J Qual Improv. 1995;21:541-548.<br />
21. Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable<br />
adverse drug events in hospitalized patients: a comparative study <strong>of</strong> intensive care <strong>and</strong><br />
general care units. Crit Care Med. 1997;25:1289-1297.<br />
22. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events: Implications for prevention. JAMA.<br />
1995;274:29-34.<br />
23. Jha AK, Kuperman GJ, Teich JM, Leape L, Shea B, Rittenberg E, et al. Identifying adverse<br />
drug events: development <strong>of</strong> a computer-based monitor <strong>and</strong> comparison with chart review<br />
<strong>and</strong> stimulated voluntary report. J Am Med Inform Assoc. 1998;5:305-314.<br />
24. Adverse drug events: the magnitude <strong>of</strong> health risk is uncertain because <strong>of</strong> limited incidence<br />
data. Washington, DC: General Accounting Office; 2000.<br />
25. G<strong>and</strong>hi TK, Burstin HR, Cook EF, Puopolo AL, Haas JS, Brennan TA, et al. Drug<br />
complications in outpatients. J Gen Intern Med. 2000;15:149-154.<br />
26. Ringold DJ, Santell JP, Schneider PJ. ASHP national survey <strong>of</strong> pharmacy practice in acute<br />
care settings: dispensing <strong>and</strong> administration–1999. Am J Health Syst Pharm.<br />
2000;57:1759-1775.<br />
27. Hanlon JT, Weinberger M, Samsa GP, Schmader KE, Uttech KM, Lewis IK, et al. A<br />
r<strong>and</strong>omized, controlled trial <strong>of</strong> a clinical pharmacist intervention to improve inappropriate<br />
prescribing in elderly outpatients with polypharmacy. Am J Med. 1996;100:428-437.<br />
28. Freeman CD, Quintiliani R, Nightingale CH. Vancomycin therapeutic drug monitoring: is<br />
it necessary? Ann Pharmacother. 1993;27:594-598.<br />
29. Cantu TG, Yamanaka-Yuen NA, Lietman PS. Serum vancomycin concentrations:<br />
reappraisal <strong>of</strong> their clinical value. Clin Infect Dis. 1994;18:533-543.<br />
30. Flather MD, Yusuf S, Kober L, Pfeffer M, Hall A, Murray G, et al. Long-term ACEinhibitor<br />
therapy in patients with heart failure or left-ventricular dysfunction: a systematic<br />
overview <strong>of</strong> data from individual patients. ACE-Inhibitor Myocardial Infarction<br />
Collaborative Group. Lancet. 2000;355:1575-1581.<br />
31. McMullin ST, Hennenfent JA, Ritchie DJ, Huey WY, Lonergan TP, Schaiff RA, et al. A<br />
prospective, r<strong>and</strong>omized trial to assess the cost impact <strong>of</strong> pharmacist-initiated<br />
interventions. Arch Intern Med. 1999;159:2306-2309.<br />
77
Chapter 8. Computer Adverse Drug Event (ADE) Detection <strong>and</strong> Alerts<br />
Tejal K. G<strong>and</strong>hi, MD, MPH<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
Adverse drug events (ADEs) occur in both inpatient <strong>and</strong> outpatient settings. 1,2 Most<br />
institutions use spontaneous incident reporting to detect adverse events in general, <strong>and</strong> ADEs in<br />
particular. Spontaneous incident reporting relies exclusively on voluntary reports from nurses,<br />
pharmacists <strong>and</strong> physicians focused on direct patient care. However, spontaneous reporting is<br />
ineffective, identifying only one in 20 ADEs. 3 Efforts to increase the frequency <strong>of</strong> spontaneous<br />
reporting have had only a minor impact.<br />
Several studies demonstrate the effectiveness <strong>of</strong> using computerized detection <strong>and</strong> alert<br />
systems (referred to as computer “monitors”) to detect ADEs. In 1991, Classen et al 4 published<br />
information about a computerized ADE monitor that was programmed to identify signals—in<br />
effect mismatches <strong>of</strong> clinical information—that suggested the presence <strong>of</strong> an ADE. The signals<br />
included sudden medication stop orders, antidote ordering, <strong>and</strong> certain abnormal laboratory<br />
values. 4 The computerized signals were then evaluated by a pharmacist who determined whether<br />
an ADE had occured. Based on the rules <strong>of</strong> this study, Jha et al developed a similar monitor that<br />
identified approximately half the ADEs identified by chart review, at much lower cost. 5<br />
Similarly, Bowman <strong>and</strong> Carlstedt used the Regenstrief <strong>Medical</strong> Record System to create a<br />
computerized inpatient ADE detection system. 6 Compared to the “gold st<strong>and</strong>ard” <strong>of</strong> chart<br />
review, the monitor had 66% sensitivity <strong>and</strong> 61% specificity, with a positive predictive value <strong>of</strong><br />
0.34. Finally, one community hospital implemented an ADE monitor with triggers that were<br />
reviewed by pharmacists who then contacted physicians to make appropriate regimen changes.<br />
This study identified opportunities to prevent patient injury at a rate <strong>of</strong> 64/1000 admissions. 7<br />
These studies <strong>and</strong> others demonstrate the potential value <strong>of</strong> computer monitors, especially when<br />
linked to effective integrated information systems. While monitors are not yet widely used, they<br />
<strong>of</strong>fer an efficient approach for monitoring the frequency <strong>of</strong> ADEs on an ongoing basis, <strong>and</strong> the<br />
Health Care Financing Administration is considering m<strong>and</strong>ating them. 8<br />
Practice Description<br />
Computerized ADE alert monitors use rule sets to search signals that suggest the<br />
presence <strong>of</strong> adverse drug events. The most frequently studied rule sets (or “triggers”) are those<br />
that search for drug names (eg, naloxone, kayexalate), drug-lab interactions (eg, heparin <strong>and</strong><br />
elevated PTT) or lab levels alone (eg, elevated digoxin levels) that frequently reflect an ADE.<br />
Simple versions can be implemented with pharmacy <strong>and</strong> laboratory data alone, although the<br />
yield <strong>and</strong> positive predictive value <strong>of</strong> signals is higher when the 2 databases are linked.<br />
Further refinements include searches for International Classification <strong>of</strong> Diseases (ICD-9)<br />
codes, <strong>and</strong> text-searching <strong>of</strong> electronic nursing bedside charting notes or outpatient notes for<br />
drug-symptom combinations (eg, medication list includes an angiotensin converting enzyme<br />
inhibitor <strong>and</strong> the patient notes mention “cough”). Although these refinements do increase the<br />
yield <strong>of</strong> monitors, they require linkage to administrative databases or electronic medical records.<br />
The information captured with computer monitors is used to alert a responsible clinician<br />
or pharmacist, who can then change therapy based on the issue in question. Systems are designed<br />
79
to alert the monitoring clinician in various ways. Alerts can go to one central location (eg,<br />
hospital pharmacist) or to individual physicians. Monitoring pharmacists typically review the<br />
alert <strong>and</strong> contact the appropriate physician if they determine that the alert has identified a true<br />
event. The alert modality also varies based on the available technology, from printed out reports,<br />
to automatic paging <strong>of</strong> covering physicians, to display <strong>of</strong> alerts on computer systems (either in<br />
results or ordering applications). It should be emphasized that computerized physician order<br />
entry (Chapter 6) is not a requirement for these monitors. Thus, a simple version <strong>of</strong> this approach<br />
could be implemented in most US hospitals.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
It is estimated that over 770,000 people are injured or die in hospitals from ADEs<br />
annually, 3,9,10 but variations in study criteria <strong>and</strong> definitions prevent precise national estimates. 11<br />
Fewer data address the epidemiology <strong>of</strong> ADEs in the outpatient setting. One recent study found<br />
an ADE rate <strong>of</strong> 3% in adult primary care outpatients, 12 while an older study reported a similar<br />
rate <strong>of</strong> 5% among ambulatory patients attending an internal medicine clinic over a 1-year<br />
period. 2<br />
Detection <strong>and</strong> alerting interventions primarily target ADEs related to the medication<br />
ordering process. One study <strong>of</strong> preventable inpatient ADEs in adults demonstrated that 56%<br />
occurred at the stage <strong>of</strong> ordering. 13 Among the 6.5 ADEs per 100 admissions in this study, 28%<br />
were judged preventable, principally by changing the systems by which drugs are ordered <strong>and</strong><br />
administered. 14 In one study <strong>of</strong> computerized ADE detection, the ADEs identified by<br />
computerized monitoring were significantly more likely to be classified as severe than those<br />
identified by chart review (51% vs. 42%, p=0.04). 5 Thus, monitors may capture a subset <strong>of</strong><br />
events with the most potential for patient injury.<br />
Injuries due to drugs have important economic consequences. Inpatients that suffer ADEs<br />
have increased lengths <strong>of</strong> stay <strong>of</strong> nearly 2 days <strong>and</strong> added hospital costs <strong>of</strong> more than $2000. 9,15<br />
Bootman has estimated the annual cost <strong>of</strong> drug-related morbidity <strong>and</strong> mortality within the United<br />
States at $76.6 billion, with the majority ($47 billion) related to hospital admissions due to drug<br />
therapy or the absence <strong>of</strong> appropriate drug therapy. 16<br />
Opportunities for Impact<br />
Unfortunately, there are no good data as to how many hospitals have integrated lab <strong>and</strong><br />
medication systems, which are required for many <strong>of</strong> the triggers used in computerized ADE<br />
monitors.<br />
Study Designs<br />
We found 5 studies <strong>of</strong> computerized ADE alert systems that were Level 3 or better (see<br />
Table 8.1). Four studies 17-20 detected potential ADEs using computerized monitoring <strong>and</strong> then<br />
alerted physicians or pharmacists about the event. One additional study 21 both alerted the<br />
monitoring clinician <strong>and</strong> made recommendations for actions relating to the suspect condition.<br />
Four studies 17-20 were in the inpatient setting <strong>and</strong> one was in the ambulatory setting. 21<br />
Study Outcomes<br />
All <strong>of</strong> the studies reported Level 1 or 2 outcomes. Level 1 outcomes were the rate <strong>of</strong><br />
ADEs 18,19 <strong>and</strong> renal impairment (as reflected by rises in creatinine). 20 Level 2 outcomes included<br />
percent <strong>of</strong> time recommended actions were taken <strong>and</strong> time to respond to an event. 17-20<br />
80
Evidence for Effectiveness <strong>of</strong> the Practice<br />
One study demonstrated significant decreases in adverse clinical outcomes with the alert<br />
systems. 18 This decrease included a surprisingly large reduction in allergic reactions (ones not<br />
previously known); it is not clear how the computer alert or decision support system could have<br />
contributed to this result. 18 The second study 19 showed no significant difference in the number <strong>of</strong><br />
adverse events in the intervention <strong>and</strong> control groups. This study <strong>and</strong> 3 others revealed<br />
significant improvements in response times concerning lab values, 17-20 <strong>and</strong> one study found a<br />
significant decrease in the risk <strong>of</strong> serious renal impairment. 20 Finally, one study demonstrated<br />
significant changes in physician behavior/modification <strong>of</strong> therapy based on alerts with<br />
recommended actions. 21 The effect sizes shown in these studies are listed in Table 8.1.<br />
Potential for Harm<br />
None <strong>of</strong> the studies discuss any potential for harm associated with the monitor <strong>and</strong> alerts.<br />
It is certainly possible that alerts could be erroneous, but it is doubtful that this would lead to any<br />
direct patient harm. As in studies <strong>of</strong> safety in other industries, one possible source <strong>of</strong> harm could<br />
be too many false positives. Large numbers <strong>of</strong> clinically insignificant warnings for patients<br />
would interfere with routine care, <strong>and</strong> might result in providers ignoring all warnings, even<br />
clinically meaningful ones.<br />
Costs <strong>and</strong> Implementation<br />
In general, implementation <strong>of</strong> alert monitors requires computer systems that can link<br />
laboratory <strong>and</strong> medication information. Integrating this information requires the creation <strong>of</strong><br />
interface between the drug <strong>and</strong> laboratory databases, with costs that will vary depending on the<br />
nature <strong>of</strong> the existing information systems. In addition, alert methods vary, with some<br />
applications directly contracting physicians (which requires further integration <strong>of</strong> coverage <strong>and</strong><br />
pager databases) <strong>and</strong> others using pharmacist intermediaries. The cost <strong>of</strong> pharmacist review <strong>of</strong><br />
triggers was less than 1 FTE per hospital in 2 studies 5,7 ; one <strong>of</strong> them reported an annual cost<br />
savings <strong>of</strong> up to 3 million dollars by reducing preventable ADEs with this alerting technique. 7<br />
Studies thus far suggest that physicians view computerized alert systems favorably.<br />
Forty-four percent <strong>of</strong> physician-respondents receiving alerts indicated that the alerts were helpful<br />
<strong>and</strong> 65% wished to continue receiving them (although these alerts went to many physicians<br />
because it was unclear who the responsible doctor was). In another study in which alerts were<br />
sent only to the responsible physician, 95% <strong>of</strong> physician-respondents were pleased to receive<br />
them. 19<br />
The systems in these studies were all “homegrown” <strong>and</strong> contained idiosyncrasies that<br />
might undermine their implementation elsewhere. Clearly it is important that systems track<br />
which physicians are responsible for which patients. In addition, all 4 <strong>of</strong> the inpatient studies<br />
were conducted at tertiary care hospitals <strong>and</strong> the outpatient study was done at clinics affiliated<br />
with a tertiary care center. The translation <strong>of</strong> this alerting approach to community settings may<br />
be difficult. One community teaching hospital has reported success in detecting opportunities to<br />
prevent injury (64/1000 admissions) using computerized detection <strong>and</strong> alerting. This report had<br />
only a Level 4 design (no control group), so it was not included in the Table. 7<br />
81
Comment<br />
Computerized real-time monitoring facilitates detection <strong>of</strong> actual <strong>and</strong> potential ADEs <strong>and</strong><br />
notification <strong>of</strong> clinicians. This in turn may aid in the prevention <strong>of</strong> ADEs or decrease the chances<br />
that ADEs will cause harm. The monitors also yield improvements in secondary measures<br />
relating to the length <strong>of</strong> time until response <strong>and</strong> the quality <strong>of</strong> response.<br />
The applications in these studies were all “homegrown.” Future applications should be<br />
evaluated <strong>and</strong> refined further. In particular, it is important to quantify the yield <strong>of</strong> collecting<br />
these data in terms <strong>of</strong> patient outcomes, since the start-up costs are significant. If monitors do<br />
lead to important clinical benefits, they should become st<strong>and</strong>ard features <strong>of</strong> commercially<br />
available hospital computer systems. As this occurs, careful attention will need to be paid to<br />
optimizing the response process.<br />
In addition, little has been done to translate these monitoring systems to the outpatient<br />
setting, largely because outpatient clinical information is <strong>of</strong>ten not computerized or resides in<br />
disparate systems. As integrated computerized outpatient records become more common, the<br />
systems developed in the inpatient setting should be translatable to the outpatient setting.<br />
82
Table 8.1. Included studies <strong>of</strong> computerized systems for ADE detection <strong>and</strong> alerts*<br />
Study Intervention Study Design,<br />
Outcomes<br />
Kuperman, 1999 19<br />
Computerized alerts to physicians<br />
via paging system<br />
McDonald, 1976 21<br />
Alerts to outpatient physicians<br />
with suggested responses to<br />
medication related events<br />
Evans, 1994 18<br />
Computerized monitor to detect<br />
ADEs (including drug/lab<br />
monitors <strong>and</strong> searches <strong>of</strong> nursing<br />
notes) <strong>and</strong> then computerized<br />
alerts to physicians<br />
Rind, 1994 20<br />
Computerized alert system to<br />
physicians about rising serum<br />
creatinine values in patients<br />
receiving potentially nephrotoxic<br />
medications<br />
Bradshaw, 1989 17<br />
Computerized alerts integrated<br />
into result review <strong>and</strong> alerts by<br />
flashing light<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 2<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 2<br />
83<br />
Results†<br />
Median time until initial treatment ordered:<br />
1 vs. 1.6 hours (p=0.003)<br />
Median time until condition resolved:<br />
8.4 vs. 8.9 hours (p=0.11)<br />
Number <strong>of</strong> adverse events:<br />
no significant difference<br />
Physicians performed recommended testing:<br />
36% vs. 11% (p
References<br />
1. Kellaway GS, McCrae E. Intensive monitoring for adverse drug effects in patients<br />
discharged from acute medical wards. N Z Med J. 1973;78:525-528.<br />
2. Hutchinson TA, Flegel KM, Kramer MS, Leduc DG, Kong HH. Frequency, severity <strong>and</strong><br />
risk factors for adverse drug reactions in adult out-patients: a prospective study. J Chronic<br />
Dis. 1986;39:533-542.<br />
3. Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal AR, Leape LL. The incident<br />
reporting system does not detect adverse drug events: a problem for quality improvement.<br />
Jt Comm J Qual Improv. 1995;21:541-548.<br />
4. Classen DC, Pestotnik SL, Evans RS, Burke JP. Computerized surveillance <strong>of</strong> adverse drug<br />
events in hospital patients. JAMA. 1991;266:2847-2851.<br />
5. Jha AK, Kuperman GJ, Teich JM, Leape L, Shea B, Rittenberg E, et al. Identifying adverse<br />
drug events: development <strong>of</strong> a computer-based monitor <strong>and</strong> comparison with chart review<br />
<strong>and</strong> stimulated voluntary report. J Am Med Inform Assoc. 1998;5:305-314.<br />
6. Bowman L, Carlstedt BC, Black CD. Incidence <strong>of</strong> adverse drug reactions in adult medical<br />
inpatients. Can J Hosp Pharm. 1994;47:209-216.<br />
7. Raschke RA, Gollihare B, Wunderlich TA, Guidry JR, Leibowitz AI, Peirce JC, et al. A<br />
computer alert system to prevent injury from adverse drug events: development <strong>and</strong><br />
evaluation in a community teaching hospital. JAMA. 1998;280:1317-20. Published erratum<br />
appears in JAMA 1999 Feb 3;281:420.<br />
8. Department <strong>of</strong> Health <strong>and</strong> Human Services. Health care financing administration. Fed<br />
Regist. 1997; 62.<br />
9. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in<br />
hospitalized patients. Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA.<br />
1997;277:301-306.<br />
10. Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable<br />
adverse drug events in hospitalized patients: a comparative study <strong>of</strong> intensive care <strong>and</strong><br />
general care units. Crit Care Med. 1997;25:1289-1297.<br />
11. Adverse drug events: the magnitude <strong>of</strong> health risk is uncertain because <strong>of</strong> limited incidence<br />
data. Washington, DC: General Accounting Office; 2000. GAO/HEHS-00-21.<br />
12. G<strong>and</strong>hi TK, Burstin HR, Cook EF, Puopolo AL, Haas JS, Brennan TA, et al. Drug<br />
complications in outpatients. J Gen Intern Med. 2000;15:149-154.<br />
13. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events. Implications for prevention. ADE<br />
Prevention Study Group. JAMA. 1995;274:29-34.<br />
14. Leape LL, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems<br />
analysis <strong>of</strong> adverse drug events. ADE Prevention Study Group. JAMA. 1995;274:35-43.<br />
15. Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA, et al. The costs <strong>of</strong><br />
adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study<br />
Group. JAMA. 1997;277:307-311.<br />
16. Johnson JA, Bootman JL. Drug-related morbidity <strong>and</strong> mortality. A cost-<strong>of</strong>-illness model.<br />
Arch Intern Med. 1995;155:1949-1956.<br />
17. Bradshaw KE, Gardner RM, Pryor TA. Development <strong>of</strong> a computerized laboratory alerting<br />
system. Comput Biomed Res. 1989;22:575-587.<br />
18. Evans RS, Pestotnik SL, Classen DC, Horn SD, Bass SB, Burke JP. Preventing adverse<br />
drug events in hospitalized patients. Ann Pharmacother. 1994;28:523-527.<br />
84
19. Kuperman GJ, Teich JM, Tanasijevic MJ, Ma'Luf N, Rittenberg E, Jha A, et al. Improving<br />
response to critical laboratory results with automation: results <strong>of</strong> a r<strong>and</strong>omized controlled<br />
trial. J Am Med Inform Assoc. 1999;6:512-522.<br />
20. Rind DM, Safran C, Phillips RS, Wang Q, Calkins DR, Delbanco TL, et al. Effect <strong>of</strong><br />
computer-based alerts on the treatment <strong>and</strong> outcomes <strong>of</strong> hospitalized patients. Arch Intern<br />
Med. 1994;154:1511-1517.<br />
21. McDonald CJ. Use <strong>of</strong> a computer to detect <strong>and</strong> respond to clinical events: its effect on<br />
clinician behavior. Ann Intern Med. 1976;84:162-167.<br />
85
Chapter 9. Protocols for High-Risk Drugs: Reducing Adverse Drug Events<br />
Related to Anticoagulants<br />
Tejal K. G<strong>and</strong>hi, MD, MPH<br />
Harvard <strong>Medical</strong> School<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
Published studies <strong>of</strong> adverse drug events <strong>and</strong> multiple case reports have consistently<br />
identified certain classes <strong>of</strong> medications as particularly serious threats to patient safety. 1-3 These<br />
“high risk” medications include concentrated electrolyte solutions such as potassium chloride,<br />
intravenous insulin, chemotherapeutic agents, intravenous opiate analgesics, <strong>and</strong> anticoagulants<br />
such as heparin <strong>and</strong> warfarin. Analyses <strong>of</strong> some <strong>of</strong> the adverse events involving these mediations<br />
have led to important recommendations regarding their administration. Examples include the use<br />
<strong>of</strong> order templates for chemotherapeutic agents, removal <strong>of</strong> intravenous electrolyte solutions<br />
from general ward stock, <strong>and</strong> protocols for reviewing the settings <strong>of</strong> intravenous pumps<br />
delivering continuous or frequent doses <strong>of</strong> opiates. 2,4,5 While these recommendations have high<br />
face validity, they have generally not been subject to formal evaluation regarding their impact in<br />
reducing the targeted adverse events. By contrast, several practices relating to the management<br />
<strong>of</strong> patients receiving anticoagulants have been evaluated quite extensively, <strong>and</strong> therefore<br />
constitute the focus <strong>of</strong> this chapter.<br />
Heparin <strong>and</strong> warfarin are medications whose use or misuse carry significant potential for<br />
injury. Subtherapeutic levels can lead to thromboembolic complications in patients with atrial<br />
fibrillation or deep venous thrombosis (DVT), while supratherapeutic levels can lead to bleeding<br />
complications. These medications are commonly involved in ADEs for a variety <strong>of</strong> reasons,<br />
including the complexity <strong>of</strong> dosing <strong>and</strong> monitoring, patient compliance, numerous drug<br />
interactions, <strong>and</strong> dietary interactions that can affect drug levels. Strategies to improve both the<br />
dosing <strong>and</strong> monitoring <strong>of</strong> these high-risk drugs have potential to reduce the associated risks <strong>of</strong><br />
bleeding or thromboembolic events.<br />
Practice Description<br />
The practices reviewed in this chapter are all intended to reduce dosing <strong>and</strong>/or<br />
monitoring errors for heparin <strong>and</strong> warfarin, as follows:<br />
• Heparin dosing protocols (“nomograms”) typically involve a st<strong>and</strong>ard initial<br />
bolus <strong>and</strong> infusion rate, instructions for when to draw the first partial<br />
thromboplastin time (PTT), <strong>and</strong> orders for dosing adjustments in response to<br />
this <strong>and</strong> subsequent values (so nurses can adjust doses automatically). In some<br />
cases, the initial bolus <strong>and</strong> infusion rates are based on patient weight.<br />
• Inpatient anticoagulation services for both heparin <strong>and</strong> warfarin (with or<br />
without dosing nomograms) typically consist <strong>of</strong> pharmacist-run services that<br />
provide daily pharmacy input on dosing <strong>and</strong> monitoring for patients on heparin<br />
<strong>and</strong>/or warfarin. (We excluded studies focusing solely on warfarin prophylaxis<br />
in orthopedic patients. 6 )<br />
87
• Outpatient anticoagulation clinics provide coordinated services for managing<br />
outpatient warfarin therapy. Services typically include anticoagulation<br />
monitoring <strong>and</strong> follow-up, warfarin dose adjustment, <strong>and</strong> patient education.<br />
These clinics are usually run by pharmacists or nurses operating with physician<br />
back-up, <strong>and</strong> sometimes following specific dosing nomograms.<br />
• <strong>Patient</strong> self-monitoring using a home finger-stick device <strong>and</strong> self-adjustment <strong>of</strong><br />
warfarin dosages using a nomogram. (The accuracy <strong>of</strong> these devices <strong>and</strong> the<br />
comparability <strong>of</strong> patients’ <strong>and</strong> pr<strong>of</strong>essional readings have been extensively<br />
evaluated. 7-11 )<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Intravenous heparin <strong>and</strong> oral warfarin are commonly used medications for cardiac disease<br />
<strong>and</strong> thromboembolism in the inpatient <strong>and</strong> outpatient settings. While in the aggregate they are<br />
highly beneficial (see Chapter 31), these drugs can have significant morbidities unless they are<br />
dosed <strong>and</strong> monitored appropriately. For example, inadequate therapeutic dosing <strong>of</strong> heparin can<br />
lead to increased length <strong>of</strong> stay <strong>and</strong> the potential for clot formation <strong>and</strong>/or propagation. 12 The<br />
risk <strong>of</strong> recurrent thromboembolism is reduced if the therapeutic effect <strong>of</strong> heparin is achieved<br />
quickly. 12 In addition, L<strong>and</strong>efeld et al 13 showed that the frequency <strong>of</strong> fatal, major, <strong>and</strong> minor<br />
bleeding during heparin therapy was twice that expected without heparin therapy. The effect<br />
with warfarin therapy was even more pronounced - approximately 5 times that expected without<br />
warfarin therapy. Consistent with this finding, anticoagulants accounted for 4% <strong>of</strong> preventable<br />
ADEs <strong>and</strong> 10% <strong>of</strong> potential ADEs in one large inpatient study. 1 Finally, careful drug monitoring<br />
in hospitals can reduce ADEs, suggesting that some events are due to inadequate monitoring <strong>of</strong><br />
therapies <strong>and</strong> doses. 14 These studies highlight the clear need for safety-related interventions with<br />
respect to both the dosing <strong>and</strong> monitoring <strong>of</strong> these high-risk drugs in order to prevent<br />
thromboembolic <strong>and</strong> bleeding complications.<br />
Opportunities for Impact<br />
The number <strong>of</strong> hospitals using weight-based heparin nomograms, or that have established<br />
anticoagulation clinics or services is unknown. Although common in some European countries, 15<br />
patient self-management <strong>of</strong> long-term anticoagulation with warfarin is unusual in the United<br />
States as many payers, including Medicare, do not currently cover the home testing technology. 15<br />
Study Designs<br />
Heparin nomograms were evaluated in one r<strong>and</strong>omized controlled trial (Level 1), 16 one<br />
prospective cohort comparison (Level 2) 17 <strong>and</strong> 4 controlled observational studies (Level 3). 18-21<br />
Two <strong>of</strong> these studies involved weight-based nomograms. 16,21 A third study involving a weightbased<br />
nomogram 22 was included with the studies <strong>of</strong> anticoagulation services (see below), as<br />
clinical pharmacists actively managed the dosing protocol. We excluded one retrospective<br />
before-after analysis <strong>of</strong> a weight-based heparin protocol for cardiac intensive care patients, 23<br />
because the method <strong>of</strong> selecting charts for review was never stated. Moreover, when the authors<br />
found an increase in the number <strong>of</strong> patients with excessive anticoagulation in the intervention<br />
group, they chose a second group <strong>of</strong> control patients (again with an unspecified selection<br />
method) for review, <strong>and</strong> in the end concluded that the difference was not significant.<br />
All studies <strong>of</strong> outpatient anticoagulation clinics have been Level 3 studies, typically<br />
retrospective before-after analyses, 22,24-28 although one study might more appropriately be<br />
88
egarded as a case-control study. 29 A comprehensive review <strong>of</strong> the literature on various forms <strong>of</strong><br />
anticoagulation management 30 did not meet the criteria for a systematic review, but referenced<br />
all <strong>of</strong> the additional studies <strong>of</strong> anticoagulation clinics that we could identify 31-36 <strong>and</strong> used<br />
quantitative methods to pool their results. We use the pooled results from this article 30 in Table<br />
9.2 in place <strong>of</strong> individual entries for each <strong>of</strong> these six Level 3 studies.<br />
Two studies evaluated the impact <strong>of</strong> a coordinated inpatient anticoagulation service<br />
(with or without nomograms for dosing). 22,37<br />
<strong>Patient</strong> self-management <strong>of</strong> warfarin therapy has been evaluated in at least 3 r<strong>and</strong>omized<br />
controlled trials 38-40 (Level 1) <strong>and</strong> one non-r<strong>and</strong>omized clinical trial. 41 Because a number <strong>of</strong><br />
higher-level studies exist, we did not include retrospective cohort analyses (Level 3) addressing<br />
this topic. 42-45<br />
Study Outcomes<br />
Most studies did not evaluate bleeding complications or had insufficient numbers <strong>of</strong><br />
patients to evaluate this outcome adequately. One recent study <strong>of</strong> an anticoagulation clinic’s<br />
adverse events 25 focused on anticoagulation as the primary outcome (Level 1), as did the review<br />
that pooled results from 6 observational studies <strong>of</strong> anticoagulation clinics. 30 As shown in Tables<br />
9.1-3, the rest <strong>of</strong> the studies reported Level 2 outcomes, consisting <strong>of</strong> various indicators <strong>of</strong> time<br />
to therapeutic anticoagulation <strong>and</strong> intensity or appropriateness <strong>of</strong> anticoagulation.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
• Heparin nomograms: As shown in Table 9.1, all studies showed a significant<br />
decrease (ie, improvement) in time to achievement <strong>of</strong> a therapeutic PTT <strong>and</strong>/or<br />
an increase in the proportion <strong>of</strong> patients in the therapeutic range.<br />
• Inpatient anticoagulation services: As shown in Table 9.2, both Level 3 studies<br />
evaluating this practice showed significant improvements in relevant measures<br />
22, 37<br />
<strong>of</strong> anticoagulation.<br />
• Outpatient anticoagulation services for warfarin (with <strong>and</strong> without dosing<br />
nomograms): the multiple Level 3 studies <strong>of</strong> this practice showed improvements<br />
in relevant measures <strong>of</strong> anticoagulation, with one exception. 28 This study took<br />
place in a semi-rural region <strong>of</strong> Engl<strong>and</strong>, <strong>and</strong> the hospital-based anticoagulation<br />
clinic was staffed mainly by junior physician trainees rotating through the<br />
clinic. The one study that focused primarily on Level 1 outcomes 25 showed<br />
significant reductions in adverse events related to under- or overanticoagulation.<br />
• <strong>Patient</strong> self-management: <strong>Patient</strong> self-management achieved superior measures<br />
<strong>of</strong> anticoagulation on one Level 1 comparison with routine care. 22,37 More<br />
impressive is that two Level 1 studies 38,46 <strong>and</strong> one Level 2 study 41 reported<br />
equivalent or superior measures <strong>of</strong> anticoagulation for self-management<br />
compared with anticoagulation clinics.<br />
Potential for Harm<br />
Heparin nomograms are primarily intended to achieve PTT values within the therapeutic<br />
range as quickly as possible. Although none <strong>of</strong> the studies showed increased bleeding as a result<br />
89
<strong>of</strong> aggressive anticoagulation, it is important to note that 4 <strong>of</strong> the 6 studies showed a significant<br />
increase in the proportion <strong>of</strong> patients with PTTs above the target range. 16,19-21<br />
Anticoagulation clinics carry the usual theoretical risk that increased fragmentation <strong>of</strong><br />
care will introduce new hazards, but no study showed any significant cause for concern.<br />
<strong>Patient</strong> self-monitoring clearly carries with it risks relating to the possibilities <strong>of</strong> patient<br />
misunderst<strong>and</strong>ing <strong>of</strong>, or non-compliance with dosing <strong>and</strong> monitoring protocols. No increases in<br />
adverse events were observed in the studies reviewed, but the patients evaluated in these studies,<br />
even if r<strong>and</strong>omized, were still chosen from a group <strong>of</strong> relatively compliant <strong>and</strong> motivated<br />
patients.<br />
Costs <strong>and</strong> Implementation<br />
For anticoagulation clinics, one study showed reduced costs <strong>of</strong> $162,058 per 100 patients<br />
annually, primarily through reductions in warfarin-related hospitalizations <strong>and</strong> emergency room<br />
visits. 25 Other studies indicate potential cost-savings due to reduced hospitalizations from<br />
anticoagulation-related adverse events, or show that the anticoagulation was revenue<br />
neutral. 19,24,29 Considering without these <strong>of</strong>fsetting potential savings, however, anticoagulant<br />
clinics <strong>of</strong>ten require institutional subsidy since pr<strong>of</strong>essional fee or laboratory payments do not<br />
fully cover costs.<br />
Heparin nomograms may increase lab costs due to more frequent monitoring, but one<br />
study calculated that lab costs were <strong>of</strong>fset by the need for fewer heparin boluses. 22<br />
For patient self-management <strong>of</strong> warfarin, one study showed that the cost <strong>of</strong> selfmonitoring<br />
was $11/international normalized ratio (INR) value <strong>and</strong> postulated that this would be<br />
cost-effective if it reduced the number <strong>of</strong> clinic visits. 39 Other studies have suggested that the<br />
capillary blood testing devices themselves 47 <strong>and</strong> the overall practice <strong>of</strong> patient self-management<br />
are cost-effective. 48,49 In the United States, the home monitoring devices sell for approximately<br />
$1000. Factoring in the price <strong>of</strong> cartridges <strong>and</strong> assuming the devices operate without requiring<br />
repair for 5 years, one source estimated an annual cost <strong>of</strong> approximately $600. 40<br />
Implementation <strong>of</strong> a heparin nonogram appears feasible, <strong>and</strong> was well received by<br />
physicians <strong>and</strong> nurses. 18 Physician/staff education about the protocols was important to its<br />
success. 23,24 One study showed a high level <strong>of</strong> physician <strong>and</strong> patient satisfaction with an<br />
anticoagulation clinic. 24 In addition, multiple studies reveal that patients who self-manage<br />
warfarin have significantly higher levels <strong>of</strong> satisfaction <strong>and</strong> experience less anxiety. 9,10,38,39<br />
90
Comment<br />
The primary purpose <strong>of</strong> heparin nomograms is the timely achievement <strong>of</strong> therapeutic<br />
anticoagulation, <strong>and</strong> their superiority in this regard (compared with routine care) has been<br />
convincingly established. While none <strong>of</strong> the studies showed adverse consequences <strong>of</strong> this focus<br />
on timely anticoagulation, the trend toward increases in excessive anticoagulation presents safety<br />
concerns. Studies powered to detect significant differences in bleeding complications in patients<br />
being managed with heparin dosing protocols may be warranted.<br />
The literature on anticoagulation clinics consists entirely <strong>of</strong> observational studies with<br />
important possible confounders. Nonetheless, with one exception 28 they are consistently shown<br />
to achieve superior measures <strong>of</strong> anticoagulation, <strong>and</strong> in one study, 25 superior clinical outcomes.<br />
Among the practices reviewed in this chapter, the literature on patient self-management is<br />
perhaps the most impressive. Three r<strong>and</strong>omized trials <strong>and</strong> one non-r<strong>and</strong>omized clinical trial<br />
show that patient control <strong>of</strong> anticoagulation is at least equivalent, if not superior, to management<br />
by usual care or an anticoagulation clinic. Additional observational studies reach the same<br />
results. 42-45 Thus, a relatively substantial literature supports patient self-management for<br />
outpatient warfarin therapy for motivated patients able to comply with the monitoring <strong>and</strong> dosing<br />
protocols. These studies clearly involved select groups <strong>of</strong> patients, 9 so that a larger r<strong>and</strong>omized<br />
trial with intention-to-treat analysis would be helpful.<br />
Many insurance carriers in the United States, including Medicare, do not currently<br />
subsidize the home testing technology or provide only partial coverage. 15 Despite the relatively<br />
high cost <strong>of</strong> the home testing devices, this practice may nonetheless be cost-effective due to<br />
reduced use <strong>of</strong> other clinical services. 48,49 A larger US study or detailed cost-effectiveness<br />
analysis appears warranted, especially given the higher level <strong>of</strong> patient satisfaction with this<br />
approach as compared with outpatient anticoagulation.<br />
91
Table 9.1. Studies focused primarily on heparin or warfarin nomograms*<br />
Study Study Design, Outcomes Results†<br />
Raschke, 1993 16<br />
Weight-based heparin<br />
nomogram for patients<br />
with venous<br />
thromboembolism or<br />
unstable angina<br />
Elliott, 1994 17<br />
Use <strong>of</strong> heparin<br />
nomogram for patients<br />
with acute proximal<br />
deep venous<br />
thrombosis<br />
Brown, 1997 21<br />
Weight-based heparin<br />
nomogram for ICU<br />
patients requiring<br />
acute anticoagulation<br />
with unfractionated<br />
heparin<br />
Cruickshank, 1991 18<br />
Heparin nomogram for<br />
patients with acute<br />
venous<br />
thromboembolism<br />
R<strong>and</strong>omized controlled trial<br />
(Level 1)<br />
Various markers <strong>of</strong> adequate<br />
anticoagulation (Level 2)<br />
Non-r<strong>and</strong>omized clinical trial<br />
(Level 2)<br />
Time to therapeutic PTT<br />
(Level 2)<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Time to therapeutic PTT<br />
(Level 2)<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Time to first therapeutic PTT,<br />
time to correct subsequent<br />
PTTs, time outside the<br />
therapeutic range (Level 2)<br />
92<br />
PTT in therapeutic range within 24<br />
hours: 97% vs. 77% (p
Table 9.1. Studies focused primarily on heparin or warfarin nomograms* (cont.)<br />
Study Study Design, Outcomes Results†<br />
Hollingsworth, 1995 19<br />
Heparin nomogram for<br />
hospitalized patients<br />
with acute venous<br />
thromboembolism<br />
Phillips, 1997 20<br />
Inpatient heparin <strong>and</strong><br />
warfarin nomograms<br />
<strong>and</strong> monitoring charts<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Primary outcome <strong>of</strong> the study<br />
was length <strong>of</strong> hospital stay<br />
(Level 3) but time to<br />
therapeutic PTT was a<br />
secondary outcome (Level 2)<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Measures <strong>of</strong> under- <strong>and</strong> overanticoagulation<br />
(Level 2)<br />
93<br />
Time to therapeutic PTT: 17.9 vs. 48.8<br />
hours (p
Table 9.2. Inpatient anticoagulation services <strong>and</strong> outpatient anticoagulation clinics*<br />
Study Study Design, Outcomes Results<br />
Ansell, 1996 30<br />
Pooled comparison <strong>of</strong><br />
anticoagulation clinics <strong>and</strong><br />
routine medical care<br />
Hamby, 2000 29<br />
<strong>Analysis</strong> <strong>of</strong> adverse events<br />
related to outpatient<br />
warfarin therapy among 395<br />
patients followed at a<br />
Veterans Affairs Hospital,<br />
with 306 enrolled in an<br />
anticoagulation clinic <strong>and</strong><br />
89 patients receiving usual<br />
care<br />
Lee, 1996 26<br />
Comparison <strong>of</strong> pharmacistmanaged<br />
anticoagulation<br />
clinic with patient receiving<br />
usual care<br />
Ellis, 1992 37<br />
Pharmacy-managed<br />
inpatient anticoagulation<br />
service (flow sheet for<br />
monitoring, but no<br />
nomogram) for monitoring<br />
patients receiving warfarin<br />
for a variety <strong>of</strong> indications<br />
Pooled results from 6 Level<br />
3 study designs comparing<br />
anticoagulation clinics with<br />
routine medical care 31-36<br />
(Level 3A)<br />
Major bleeding <strong>and</strong><br />
thromboembolic events<br />
(Level 1)<br />
Case-control study<br />
(Level 3)<br />
Adverse events related to<br />
under- or overanticoagulation<br />
(Level 1)<br />
Retrospective cohort<br />
comparison (Level 3)<br />
Hospital admissions related<br />
to under- or overanticoagulation<br />
– ie,<br />
thromboembolic or bleeding<br />
events (Level 1) †<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Anticoagulation “stability”<br />
at discharge <strong>and</strong> odds <strong>of</strong><br />
therapeutic anticoagulation<br />
at first outpatient visit<br />
(Level 2)<br />
94<br />
Major bleeding events per patientyear:<br />
anticoagulation clinic, 0.028<br />
(95% CI: 0-0.069) vs. routine care,<br />
0.109 (95% CI: 0.043-0.268)<br />
Thromboembolic events per<br />
patient-year: anticoagulation<br />
clinic, 0.024 (95% CI: 0-0.08) vs.<br />
routine care, 0.162 (95% CI:<br />
0.062-0.486)<br />
Among the 12 patients with<br />
preventable adverse events related<br />
to anticoagulation, 8 were not<br />
enrolled in the anticoagulation<br />
clinic<br />
<strong>Patient</strong>s receiving usual care had<br />
20 times the relative risk (95% CI:<br />
6-62) <strong>of</strong> an adverse event<br />
compared with patients in the<br />
anticoagulation clinic.<br />
<strong>Patient</strong>s in anticoagulation clinic<br />
had non-significant reductions in<br />
hospital admissions related to<br />
thromboembolic or bleeding<br />
events compared with control<br />
group ‡<br />
<strong>Patient</strong>s receiving the intervention<br />
were more likely to have PT<br />
“stability” at discharge: 61.5% vs.<br />
42.3% (p=0.02)<br />
Odds <strong>of</strong> having therapeutic PT at<br />
first outpatient clinic visit with<br />
intervention: OR 5.4 (95% CI:<br />
1.87-15.86)
Table 9.2. Inpatient anticoagulation services <strong>and</strong> outpatient anticoagulation clinics* (cont.)<br />
Study Study Design, Outcomes Results<br />
Gaughan, 2000 24<br />
Anticoagulation clinic for<br />
outpatients receiving<br />
warfarin for atrial<br />
fibrillation (managed by<br />
nurse practitioner using<br />
warfarin dosing nomogram)<br />
Radley, 1995 27<br />
Performance <strong>of</strong> pharmacistrun<br />
hospital-based<br />
outpatient anticoagulation<br />
clinic in Engl<strong>and</strong> compared<br />
with historical control<br />
(management by rotating<br />
physician trainees)<br />
Rivey, 1993 22<br />
Pharmacy-managed<br />
inpatient anticoagulation<br />
service (using weight-based<br />
heparin protocol) for<br />
medicine inpatients<br />
compared with older fixeddose<br />
protocol without any<br />
active management by<br />
pharmacists<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Percentage <strong>of</strong> patients in the<br />
desired range for<br />
anticoagulation (Level 2)<br />
was evaluated as a<br />
secondary outcome<br />
Retrospective before-after<br />
analysis (Level 3)<br />
Proportions <strong>of</strong> INR<br />
measurements “in” or “out”<br />
<strong>of</strong> the therapeutic range<br />
Before-after analysis<br />
(Level 3)<br />
Time to therapeutic PTT<br />
(Level 2)<br />
95<br />
Minor increase in percentage <strong>of</strong><br />
patients with INR in desired range:<br />
53.7% vs. 49.1% (p
Table 9.3. Outpatient self-management using home testing devices <strong>and</strong> dosing<br />
nomograms*<br />
Study Study Design, Outcomes Results<br />
Cromheecke, 2000 38<br />
Oral anticoagulation selfmanagement<br />
with home<br />
monitoring <strong>and</strong> dose<br />
adjustment compared with<br />
anticoagulation clinic<br />
(Netherl<strong>and</strong>s)<br />
Sawicki, 1999 39<br />
Oral anticoagulation selfmanagement<br />
with home<br />
monitoring <strong>and</strong> dose<br />
adjustment compared with<br />
routine care (Germany)<br />
White, 1989 40<br />
Oral anticoagulation selfmanagement<br />
with home<br />
monitoring <strong>and</strong> dose<br />
adjustment compared with<br />
anticoagulation clinic<br />
(United States)<br />
Watzke, 2000 41<br />
Self-management<br />
compared with<br />
anticoagulation clinic<br />
(Austria)<br />
R<strong>and</strong>omized trial with<br />
crossover comparison<br />
(Level 1)<br />
Adequacy <strong>of</strong><br />
anticoagulation (Level 2)<br />
Single blind, multicenter<br />
r<strong>and</strong>omized controlled<br />
trial (Level 1)<br />
Adequacy <strong>of</strong><br />
anticoagulation (Level 2)<br />
R<strong>and</strong>omized prospective<br />
comparison (Level 1)<br />
Adequacy <strong>of</strong><br />
anticoagulation (Level 2)<br />
Prospective cohort<br />
comparison (Level 2)<br />
Various measures <strong>of</strong><br />
adequacy <strong>of</strong><br />
anticoagulation (Level 2)<br />
* INR indicates international normalized ratio.<br />
96<br />
Percent <strong>of</strong> self-managed<br />
measurements within 0.5 INR units <strong>of</strong><br />
therapeutic target did not differ (55%<br />
vs. 49%, p=0.06). However, 29<br />
patients (60%) during selfmanagement<br />
spent >50% <strong>of</strong> time in<br />
target range, compared with 25 (52%)<br />
during clinic management (p
References<br />
1. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events. Implications for prevention. ADE<br />
Prevention Study Group. JAMA. 1995;274:29-34.<br />
2. Cohen MR, Anderson RW, Attilio RM, Green L, Muller RJ, Pruemer JM. Preventing<br />
medication errors in cancer chemotherapy. Am J Health Syst Pharm. 1996;53:737-746.<br />
3. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, et al. Incidence<br />
<strong>of</strong> adverse events <strong>and</strong> negligence in hospitalized patients. Results <strong>of</strong> the Harvard <strong>Medical</strong><br />
Practice Study I. N Engl J Med. 1991;324:370-376.<br />
4. Leape LL, Kabcenell AI, Berwick DM. Reducing Adverse Drug Events. Boston: Institute<br />
for Healthcare Improvement; 1998.<br />
5. Massachusetts Coalition for the Prevention <strong>of</strong> <strong>Medical</strong> Errors. MHA best practice<br />
recommendations to reduce medication errors. Available at:<br />
www.mhalink.org/mcpme/mcpme_publications.htm. Accessed, 2001.<br />
6. Rivey MP, Wood RD, Allington DR, Stratton TP, Erickson CC, Stenson TA. Pharmacymanaged<br />
protocol for warfarin use in orthopedic surgery patients. Am J Health Syst Pharm.<br />
1995;52:1310-1316.<br />
7. Ansell J, Holden A, Knapic N. <strong>Patient</strong> self-management <strong>of</strong> oral anticoagulation guided by<br />
capillary (fingerstick) whole blood prothrombin times. Arch Intern Med. 1989;149:2509-<br />
2511.<br />
8. Hasenkam JM, Knudsen L, Kimose HH, Gronnesby H, Attermann J, Andersen NT, et al.<br />
Practicability <strong>of</strong> patient self-testing <strong>of</strong> oral anticoagulant therapy by the international<br />
normalized ratio (INR) using a portable whole blood monitor. A pilot investigation.<br />
Thromb Res. 1997;85:77-82.<br />
9. Sunderji R, Campbell L, Shalansky K, Fung A, Carter C, Gin K. Outpatient selfmanagement<br />
<strong>of</strong> warfarin therapy: a pilot study. Pharmacotherapy. 1999;19:787-793.<br />
10. Point-<strong>of</strong>-care prothrombin time measurement for pr<strong>of</strong>essional <strong>and</strong> patient self-testing use.<br />
A multicenter clinical experience. Oral Anticoagulation Monitoring Study Group. Am J<br />
Clin Pathol. 2001;115:288-296.<br />
11. Prothrombin measurement using a patient self-testing system. Oral Anticoagulation<br />
Monitoring Study Group. Am J Clin Pathol. 2001;115:280-287.<br />
12. Hull RD, Raskob GE, Hirsh J, Jay RM, Leclerc JR, Geerts WH, et al. Continuous<br />
intravenous heparin compared with intermittent subcutaneous heparin in the initial<br />
treatment <strong>of</strong> proximal-vein thrombosis. N Engl J Med. 1986;315:1109-1114.<br />
13. L<strong>and</strong>efeld CS, Beyth RJ. Anticoagulant-related bleeding: clinical epidemiology, prediction,<br />
<strong>and</strong> prevention. Am J Med. 1993;95:315-328.<br />
14. Evans RS, Pestotnik SL, Classen DC, Horn SD, Bass SB, Burke JP. Preventing adverse<br />
drug events in hospitalized patients. Ann Pharmacother. 1994;28:523-527.<br />
15. Becker D. Commentary on "Self-management <strong>of</strong> long-term oral anticoagulation was as<br />
effective as specialist anticoagulation clinic management.". ACP Journal Club.<br />
2001;134:62.<br />
16. Raschke RA, Reilly BM, Guidry JR, Fontana JR, Srinivas S. The weight-based heparin<br />
dosing nomogram compared with a "st<strong>and</strong>ard care" nomogram. A r<strong>and</strong>omized controlled<br />
trial. Ann Intern Med. 1993;119:874-881.<br />
97
17. Elliott CG, Hiltunen SJ, Suchyta M, Hull RD, Raskob GE, Pineo GF, et al. Physicianguided<br />
treatment compared with a heparin protocol for deep vein thrombosis. Arch Intern<br />
Med. 1994;154:999-1004.<br />
18. Cruickshank MK, Levine MN, Hirsh J, Roberts R, Siguenza M. A st<strong>and</strong>ard heparin<br />
nomogram for the management <strong>of</strong> heparin therapy. Arch Intern Med. 1991;151:333-337.<br />
19. Hollingsworth JA, Rowe BH, Brisebois FJ, Thompson PR, Fabris LM. The successful<br />
application <strong>of</strong> a heparin nomogram in a community hospital. Arch Intern Med.<br />
1995;155:2095-2100.<br />
20. Phillips WS, Smith J, Greaves M, Preston FE, Channer KS. An evaluation <strong>and</strong><br />
improvement program for inpatient anticoagulant control. Thromb Haemost. 1997;77:283-<br />
288.<br />
21. Brown G, Dodek P. An evaluation <strong>of</strong> empiric vs. nomogram-based dosing <strong>of</strong> heparin in an<br />
intensive care unit. Crit Care Med. 1997;25:1534-1538.<br />
22. Rivey MP, Peterson JP. Pharmacy-managed, weight-based heparin protocol. Am J Hosp<br />
Pharm. 1993;50:279-284.<br />
23. Nemeth JS, Marxen TL, Piltz GW. Weight-based protocol for improving heparin therapy.<br />
Am J Health Syst Pharm. 1996;53:1164-1166.<br />
24. Gaughan GL, Dolan C, Wilk-Rivard E, Geary G, Libbey R, Gilman MA, et al. Improving<br />
management <strong>of</strong> atrial fibrillation <strong>and</strong> anticoagulation in a community hospital. Jt Comm J<br />
Qual Improv. 2000;26:18-28.<br />
25. Chiquette E, Amato MG, Bussey HI. Comparison <strong>of</strong> an anticoagulation clinic with usual<br />
medical care: anticoagulation control, patient outcomes, <strong>and</strong> health care costs. Arch Intern<br />
Med. 1998;158:1641-1647.<br />
26. Lee YP, Schommer JC. Effect <strong>of</strong> a pharmacist-managed anticoagulation clinic on warfarinrelated<br />
hospital readmissions. Am J Health Syst Pharm. 1996;53:1580-1583.<br />
27. Radley AS, Hall J, Farrow M, Carey PJ. Evaluation <strong>of</strong> anticoagulant control in a<br />
pharmacist operated anticoagulant clinic. J Clin Pathol. 1995;48:545-547.<br />
28. Pell JP, McIver B, Stuart P, Malone DN, Alcock J. Comparison <strong>of</strong> anticoagulant control<br />
among patients attending general practice <strong>and</strong> a hospital anticoagulant clinic. Br J Gen<br />
Pract. 1993;43:152-154.<br />
29. Hamby L, Weeks WB, Malikowski C. Complications <strong>of</strong> warfarin therapy: causes, costs,<br />
<strong>and</strong> the role <strong>of</strong> the anticoagulation clinic. Eff Clin Pract. 2000;3:179-184.<br />
30. Ansell JE, Hughes R. Evolving models <strong>of</strong> warfarin management: anticoagulation clinics,<br />
patient self-monitoring, <strong>and</strong> patient self-management. Am Heart J. 1996;132:1095-1100.<br />
31. Bussey HI, Rospond RM, Qu<strong>and</strong>t CM, Clark GM. The safety <strong>and</strong> effectiveness <strong>of</strong> longterm<br />
warfarin therapy in an anticoagulation clinic. Pharmacotherapy. 1989;9:214-219.<br />
32. Wilt VM, Gums JG, Ahmed OI, Moore LM. Outcome analysis <strong>of</strong> a pharmacist-managed<br />
anticoagulation service. Pharmacotherapy. 1995;15:732-739.<br />
33. Cortelazzo S, Finazzi G, Viero P, Galli M, Remuzzi A, Parenzan L, et al. Thrombotic <strong>and</strong><br />
hemorrhagic complications in patients with mechanical heart valve prosthesis attending an<br />
anticoagulation clinic. Thromb Haemost. 1993;69:316-320.<br />
34. Garabedian-Ruffalo SM, Gray DR, Sax MJ, Ruffalo RL. Retrospective evaluation <strong>of</strong> a<br />
pharmacist-managed warfarin anticoagulation clinic. Am J Hosp Pharm. 1985;42:304-308.<br />
35. Hamilton GM, Childers RW, Silverstein MD. Does clinic management <strong>of</strong> anticoagulation<br />
improve the outcome <strong>of</strong> prosthetic valve patients? Clin res. 1985;33:832A.<br />
36. Cohen IA, Hutchison TA, Kirking DM, Shue ME. Evaluation <strong>of</strong> a pharmacist-managed<br />
anticoagulation clinic. J Clin Hosp Pharm. 1985;10:167-175.<br />
98
37. Ellis RF, Stephens MA, Sharp GB. Evaluation <strong>of</strong> a pharmacy-managed warfarinmonitoring<br />
service to coordinate inpatient <strong>and</strong> outpatient therapy. Am J Hosp Pharm.<br />
1992;49:387-394.<br />
38. Cromheecke ME, Levi M, Colly LP, de Mol BJ, Prins MH, Hutten BA, et al. Oral<br />
anticoagulation self-management <strong>and</strong> management by a specialist anticoagulation clinic: a<br />
r<strong>and</strong>omised cross-over comparison. Lancet. 2000;356:97-102.<br />
39. Sawicki PT. A structured teaching <strong>and</strong> self-management program for patients receiving<br />
oral anticoagulation: a r<strong>and</strong>omized controlled trial. Working Group for the Study <strong>of</strong> <strong>Patient</strong><br />
Self-Management <strong>of</strong> Oral Anticoagulation. JAMA. 1999; 281:145-150.<br />
40. White RH, Becker DM, Gunther-Maher MG. Outpatient use <strong>of</strong> a portable international<br />
normalized ratio/prothrombin time monitor. South Med J. 1994;87:206-210.<br />
41. Watzke HH, Forberg E, Svolba G, Jimenez-Boj E, Krinninger B. A prospective controlled<br />
trial comparing weekly self-testing <strong>and</strong> self-dosing with the st<strong>and</strong>ard management <strong>of</strong><br />
patients on stable oral anticoagulation. Thromb Haemost. 2000;83:661-665.<br />
42. Christensen TD, Attermann J, Pilegaard HK, Andersen NT, Maegaard M, Hasenkam JM.<br />
Self-management <strong>of</strong> oral anticoagulant therapy for mechanical heart valve patients. Sc<strong>and</strong><br />
Cardiovasc J. 2001;35:107-113.<br />
43. Heidinger KS, Bernardo A, Taborski U, Muller-Berghaus G. Clinical outcome <strong>of</strong> selfmanagement<br />
<strong>of</strong> oral anticoagulation in patients with atrial fibrillation or deep vein<br />
thrombosis. Thromb Res. 2000;98:287-293.<br />
44. Hasenkam JM, Kimose HH, Knudsen L, Gronnesby H, Halborg J, Christensen TD, et al.<br />
Self management <strong>of</strong> oral anticoagulant therapy after heart valve replacement. Eur J<br />
Cardiothorac Surg. 1997;11:935-942.<br />
45. Ansell JE, Patel N, Ostrovsky D, Nozzolillo E, Peterson AM, Fish L. Long-term patient<br />
self-management <strong>of</strong> oral anticoagulation. Arch Intern Med. 1995;155:2185-2189.<br />
46. White RH, McCurdy SA, von Marensdorff H, Woodruff DE, Jr., Leftg<strong>of</strong>f L. Home<br />
prothrombin time monitoring after the initiation <strong>of</strong> warfarin therapy. A r<strong>and</strong>omized,<br />
prospective study. Ann Intern Med. 1989;111:730-737.<br />
47. Ansell JE, Hamke AK, Holden A, Knapic N. Cost effectiveness <strong>of</strong> monitoring warfarin<br />
therapy using st<strong>and</strong>ard versus capillary prothrombin times. Am J Clin Pathol. 1989;91:587-<br />
589.<br />
48. Taborski U, Wittstamm FJ, Bernardo A. Cost-effectiveness <strong>of</strong> self-managed anticoagulant<br />
therapy in Germany. Semin Thromb Hemost. 1999;25:103-107.<br />
49. Lafata JE, Martin SA, Kaatz S, Ward RE. Anticoagulation clinics <strong>and</strong> patient self-testing<br />
for patients on chronic warfarin therapy: A cost-effectiveness analysis. J Thromb<br />
Thrombolysis. 2000;9(Suppl 1):S13-S19.<br />
99
100
Chapter 10. Unit-Dose Drug Distribution Systems<br />
Michael D. Murray, PharmD, MPH<br />
Purdue University School <strong>of</strong> Pharmacy<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Medication errors are especially likely when health pr<strong>of</strong>essionals engage in multiple<br />
tasks within a short time span. This situation occurs repeatedly in hospitals when pharmacists<br />
<strong>and</strong> technicians load unit-dose carts, <strong>and</strong> when nurses administer medications. Unit-dose carts<br />
are prepared daily, <strong>of</strong>ten manually, by technicians <strong>and</strong> then checked by pharmacists. These carts,<br />
containing thous<strong>and</strong>s <strong>of</strong> patient-specific dosages <strong>of</strong> drugs, are sent to the wards daily, for nurses<br />
to administer medications to patients. Dosing frequencies vary widely, ranging from regular<br />
intervals around the clock to “stat” doses given to control acute pain or other symptoms.<br />
Medication administration alone is an enormous task for nurses, <strong>and</strong> one in which they are<br />
repeatedly interrupted. It is not surprising that the administration phase <strong>of</strong> medication use is<br />
particularly vulnerable to error. 1<br />
Unit-dose dispensing <strong>of</strong> medication was developed in the 1960s to support nurses in<br />
medication administration <strong>and</strong> reduce the waste <strong>of</strong> increasingly expensive medications. Most <strong>of</strong><br />
the investigations <strong>of</strong> medication errors <strong>and</strong> unit-dose dispensing took place from 1970 to 1976.<br />
Now, unit-dose dispensing <strong>of</strong> medications is a st<strong>and</strong>ard <strong>of</strong> practice at hospitals in the United<br />
States. This chapter reviews the evidence supporting this practice.<br />
Practice Description<br />
In unit-dose dispensing, medication is dispensed in a package that is ready to administer<br />
to the patient. 2 It can be used for medications administered by any route, but oral, parenteral, <strong>and</strong><br />
respiratory routes are especially common. When unit-dose dispensing first began, hospital<br />
pharmacies equipped themselves with machines that packaged <strong>and</strong> labeled tablets <strong>and</strong> capsules,<br />
one pill per package. They also purchased equipment for packaging liquids in unit-doses. As the<br />
popularity <strong>of</strong> this packaging increased, the pharmaceutical industry began prepackaging pills in<br />
unit-<strong>of</strong>-use form. Many hospitals now purchase prepackaged unit-dose medications. However, it<br />
is still common for hospital pharmacies to purchase bulk supplies <strong>of</strong> tablets <strong>and</strong> capsules from<br />
manufacturers <strong>and</strong> repackage them in the central pharmacy into unit-dose packages. 2 It is<br />
important to note that hospitals vary in the proportion <strong>of</strong> their wards covered by a unit-dose<br />
system.<br />
There are many variations <strong>of</strong> unit-dose dispensing. As just one example, when physicians<br />
write orders for inpatients, these orders are sent to the central pharmacy (by pharmacists, nurses,<br />
other personnel, or computer). Pharmacists verify these orders <strong>and</strong> technicians place drugs in<br />
unit-dose carts. The carts have drawers in which each patient’s medications are placed by<br />
pharmacy technicians—one drawer for each patient. The drawers are labeled with the patient’s<br />
name, ward, room, <strong>and</strong> bed number. Before the carts are transported to the wards, pharmacists<br />
check each drawer’s medications for accuracy. Sections <strong>of</strong> each cart containing all medication<br />
drawers for an entire nursing unit <strong>of</strong>ten slide out <strong>and</strong> can be inserted into wheeled medication<br />
carts used by nurses during their medication administration cycles. A medication administration<br />
recording form sits on top <strong>of</strong> the cart <strong>and</strong> is used by the nurse to check-<strong>of</strong>f <strong>and</strong> initial the time <strong>of</strong><br />
each administration <strong>of</strong> each medication. The next day, the carts are retrieved from the wards <strong>and</strong><br />
101
eplaced by a fresh <strong>and</strong> updated medication supply. Medications that have been returned to the<br />
central pharmacy are credited to the patient’s account.<br />
A 1999 national survey <strong>of</strong> drug dispensing <strong>and</strong> administration practices indicated that<br />
three-fourths <strong>of</strong> responding hospitals had centralized pharmacies, 77% <strong>of</strong> which were not<br />
automated. 2 Larger hospitals <strong>and</strong> those affiliated with medical schools were more likely to have<br />
some component <strong>of</strong> decentralized pharmacy services. About half <strong>of</strong> the surveyed hospitals<br />
reported drug distribution “systems” that bypassed the pharmacy, including hospitals that<br />
reported using floor stocks, borrowing other patients’ medications, <strong>and</strong> hidden drug supplies.<br />
Studies <strong>of</strong>ten compare unit-dose dispensing to a ward stock system. In this system, nurses<br />
order drugs in bulk supplies from the pharmacy; the drugs are stored in a medication room on the<br />
ward. Nurses prepare medication cups for each patient during medication administration cycles.<br />
The correct number <strong>of</strong> pills must be taken out <strong>of</strong> the correct medication container for each cycle<br />
<strong>and</strong> taken to the patient for administration. Liquids must be poured by the nurse from the<br />
appropriate bottle <strong>and</strong> each dose carefully measured. Nurses are responsible for any necessary<br />
labeling. Any medications taken from stock bottles <strong>and</strong> not administered to patients are generally<br />
disposed <strong>of</strong>.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The targets <strong>of</strong> the safety problem for unit-dosing are drug dispensing 3 <strong>and</strong><br />
administration. 4,5 Improving these stages probably carries the greatest opportunity to reduce<br />
medication errors.<br />
Bates et al 6 identified 530 medical errors in 10,070 written orders for drugs (5.3<br />
errors/100 orders) on 3 medical units observed for 51 days. Of the 530 errors, 5 (0.9%) resulted<br />
in an adverse drug event. The most common reason for an error was a missing dose <strong>of</strong><br />
medication, which occurred in 53% <strong>of</strong> orders. In a systems analysis <strong>of</strong> 334 errors causing 264<br />
adverse drug events over 6 months in 2 tertiary care hospitals, 130 errors (39%) resulted from<br />
physician ordering, 40 (12%) involved transcription <strong>and</strong> verification, 38 (11%) reflected<br />
problems with pharmacy dispensing, <strong>and</strong> 126 (38%) were from nursing administration. 4 In other<br />
words, 164 (49%) <strong>of</strong> the errors in the above-cited study 4 were in the dispensing <strong>and</strong><br />
administration stages. In further research, the investigators found that errors resulting in<br />
preventable adverse drug events were more than likely to be those in the administration stage<br />
(34%) than those in the dispensing stage (4%) <strong>of</strong> the medication use process. 1<br />
Opportunities for Impact<br />
Because unit-dose dispensing now constitutes a st<strong>and</strong>ard for approval by the Joint<br />
Commission for Accreditation <strong>of</strong> Healthcare Organizations (JCAHO) 7,8 <strong>and</strong> is closely linked to<br />
the increasingly common use <strong>of</strong> automated dispensing devices (see Chapter 11), there is likely<br />
little opportunity for further implementation <strong>of</strong> this practice in US hospitals. In a 1994 survey <strong>of</strong><br />
pharmacy directors, 92% <strong>of</strong> acute care hospitals reported using unit-dose dispensing. 7 Use <strong>of</strong><br />
unit-dose dispensing is extremely common on general medical <strong>and</strong> surgical wards, but less so in<br />
other locations such as intensive care units, operating rooms, <strong>and</strong> emergency departments. In<br />
these areas, bulk medication stock systems are still found. In a 1999 survey <strong>of</strong> pharmacy<br />
directors, 80% reported that unit-dose dispensing was used for 75% or more <strong>of</strong> oral doses, <strong>and</strong><br />
52% <strong>of</strong> injectable medications dispensed in their hospitals. 9<br />
102
Study Designs<br />
The 5 studies meeting inclusion criteria for this review (Table 10.1) included 4 Level 3<br />
studies, <strong>and</strong> one prospectively designed before-after observation (Level 2 design). We excluded<br />
the following studies for the reasons outlined:<br />
• Read 10 conducted a study <strong>of</strong> a unit-dose system applied to respiratory therapy<br />
solutions that was focused primarily on measures <strong>of</strong> efficiency.<br />
•<br />
Volume/concentration errors in preparing respiratory solutions were also<br />
discussed, but the method <strong>of</strong> detection was not stated nor were specific error<br />
data provided for the study period.<br />
Reitberg et al 11 conducted a prospective before-after study with a concurrent<br />
comparison floor as a control (Level 2 design) in which medication errors<br />
represented the main outcomes (Level 2). During period 1 when both wards<br />
used the ward stock system, total errors were 4.7% on the ward that remained<br />
on ward stock <strong>and</strong> 34.2% on the ward that used the modified dispensing<br />
system. During period 2, the ward stock system had 5.1% error, while the<br />
intervention ward (unit-dose) exhibited a significantly increased 18% error<br />
rate. Excluding administration-time errors resulted in error rates <strong>of</strong> 5.1%<br />
(conventional system) <strong>and</strong> 4.8% (unit-dose), which are not significantly<br />
different. The greater number <strong>of</strong> errors on the modified dispensing ward<br />
before <strong>and</strong> after the intervention were attributed by the authors to factors other<br />
than the dispensing systems. Because <strong>of</strong> this problem, <strong>and</strong> the fact that this<br />
study took place in a skilled nursing facility rather than an acute care hospital,<br />
we excluded this study.<br />
• Shultz et al 12 conducted a prospective before-after study (Level 2) involving a<br />
somewhat complicated comparison. In the baseline period, approximately<br />
50% <strong>of</strong> wards used a unit-dose system. During the study period, some wards<br />
switched to a unit-dose system in which nurses still administered the<br />
medications, while other wards adopted an experimental system, in which<br />
pharmacists <strong>and</strong> pharmacy technicians h<strong>and</strong>led all aspects <strong>of</strong> the mediation<br />
dispensing <strong>and</strong> administration process. The authors report an error rate <strong>of</strong><br />
0.64% for complete unit-dose plus pharmacy technician medication system<br />
compared to 5.3% in the more conventional unit-dose system in which nurses<br />
continue to administer medications (p
Lastly, we were unable to obtain one <strong>of</strong> the original studies <strong>of</strong> the unit-dose system<br />
within the timeline <strong>of</strong> the project. The data from this study appears to have been published only<br />
as a technical report. 15 Other publications related to this study <strong>and</strong> which we were able to<br />
obtain 16-18 did not provide sufficient detail on the aspects <strong>of</strong> the study relating to medication<br />
errors to permit abstraction or inclusion in this chapter. The published reports <strong>of</strong> a study<br />
conducted in a private hospital similarly did not contain sufficient detail about the study methods<br />
or results to permit abstraction <strong>and</strong> inclusion. 19,20<br />
Study Outcomes<br />
Studies reported errors measured by direct observation using a methodology that was first<br />
described by Barker. 21 All <strong>of</strong> these studies involved Level 2 outcomes.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Though the practice <strong>of</strong> unit-dose dispensing is generally well accepted <strong>and</strong> has been<br />
widely implemented, the evidence for its effectiveness is modest. Most <strong>of</strong> the published studies<br />
reported reductions in medication errors <strong>of</strong> omission <strong>and</strong> commission with unit-dose dispensing<br />
compared with alternative dispensing systems such as ward stock systems. One exception was<br />
the international study by Dean, 22 which compared a United States hospital using unit-dose <strong>and</strong><br />
transcription <strong>of</strong> medication orders with a United Kingdom hospital using ward stock <strong>and</strong> no<br />
transcription <strong>of</strong> orders. The results <strong>of</strong> this study indicated that the United Kingdom hospital had<br />
less than half the errors <strong>of</strong> the United States hospital. The study groups were <strong>of</strong>ten difficult to<br />
compare because <strong>of</strong> differing cultures, hospitals, or nursing care units. Moreover, cointerventions<br />
undoubtedly confounded the results. Each study is summarized in Table 10.1.<br />
Potential for Harm<br />
Unit-dosing shifts the effort <strong>and</strong> distraction <strong>of</strong> medication processing, with its potential<br />
for harm, from the nursing ward to central pharmacy. It increases the amount <strong>of</strong> time nurses have<br />
to do other tasks but increases the volume <strong>of</strong> work within the pharmacy. Like the nursing units,<br />
central pharmacies have their own distractions that are <strong>of</strong>ten heightened by the unit-dose<br />
dispensing process itself, <strong>and</strong> errors do occur. 3<br />
Overall, unit-dose appears to have little potential for harm. The results <strong>of</strong> most <strong>of</strong> the<br />
Level 2 observational studies seem to indicate that it is safer than other forms <strong>of</strong> institutional<br />
dispensing. However, the definitive study to determine the extent <strong>of</strong> harm has not yet been<br />
conducted.<br />
A major advantage <strong>of</strong> unit-dose dispensing is that it brings pharmacists into the<br />
medication use process at another point to reduce error. 3 Yet, as pointed out in the Practice<br />
Description above, about half <strong>of</strong> the hospitals in a national survey bypass pharmacy involvement<br />
by using floor stock, borrowing patients’ medications, <strong>and</strong> hiding medication supplies. 2<br />
Costs <strong>and</strong> Implementation<br />
The cost considerations <strong>of</strong> unit-dose dispensing are mainly a trade-<strong>of</strong>f between pharmacy<br />
<strong>and</strong> nursing personnel. The pharmacy personnel involved are mainly technicians who load unitdose<br />
ward carts for the pharmacists to check <strong>and</strong> may package some medications that are<br />
commercially unavailable in unit-dose form. The pharmacist must verify that the correct<br />
medication <strong>and</strong> dosage <strong>of</strong> each medication is sent to the ward for the nurse to administer.<br />
Nursing time to maintain a drug inventory is reduced, allowing more time for other nursing<br />
activities. A variable cost <strong>of</strong> unit-dose dispensing is the cost <strong>of</strong> equipment <strong>and</strong> supplies to those<br />
104
hospitals that wish to do much <strong>of</strong> the packaging themselves instead <strong>of</strong> purchasing medications<br />
pre-packaged as unit-doses.<br />
Comment<br />
The studies evaluating the practice <strong>of</strong> unit-dosing contain important methodologic<br />
problems <strong>and</strong>, although yielding somewhat heterogeneous results, are overall relatively<br />
consistent in showing a positive impact on error reduction. In contrast to other practices related<br />
to medication use, none <strong>of</strong> the studies evaluated Level 1 outcomes, such as actual adverse drug<br />
events. Nonetheless, unit-dose dispensing or some form <strong>of</strong> automated dispensing <strong>of</strong> unit-doses<br />
(see Chapter 11) has become ubiquitous in American hospitals <strong>and</strong> a st<strong>and</strong>ard <strong>of</strong> care in the<br />
delivery <strong>of</strong> pharmacy services. Consequently, it is unlikely that more rigorous studies could now<br />
be conducted.<br />
105
Table 10.1. Studies evaluating the impact <strong>of</strong> unit-dose dispensing on medication errors*<br />
Study Study Design, Outcomes† Results: Error Rates (95% CI)<br />
Hynniman,<br />
1970 23<br />
Means,<br />
1975 13<br />
Simborg,<br />
1975 14 ‡<br />
Schnell,<br />
1976 24<br />
Dean,<br />
1995 22<br />
Taxis,<br />
1998 25<br />
Cross-sectional comparison between<br />
study hospital <strong>and</strong> non-r<strong>and</strong>omly<br />
selected “comparison” hospitals (Level<br />
3)<br />
Errors <strong>of</strong> commission <strong>and</strong> omission<br />
(Level 2) among doses ordered<br />
Cross-sectional comparison <strong>of</strong> 2 wards<br />
within a single hospital over a 60-day<br />
period (Level 3)<br />
Errors <strong>of</strong> commission (Level 2) among<br />
doses administered during r<strong>and</strong>omly<br />
chosen observation periods<br />
Prospective before-after study (Level<br />
2) at four Canadian hospitals<br />
Errors observed during medication<br />
preparation <strong>and</strong> administration (Level<br />
2)<br />
Cross-sectional comparison (Level 3) <strong>of</strong><br />
US <strong>and</strong> UK hospitals with different<br />
pharmacy distribution systems<br />
Errors observed during medication<br />
administration (Level 2)<br />
Cross-sectional comparison (Level 3) <strong>of</strong><br />
2 hospitals in Germany <strong>and</strong> one hospital<br />
in the UK<br />
Errors observed during medication<br />
administration<br />
106<br />
Unit-dose system: 3.5% (3.1-4.0%)<br />
Conventional distribution systems at 4 hospitals:<br />
8.3% (7.1-9.7%)<br />
9.9% (8.0-12.2%)<br />
11.4% (9.9-13.2%)<br />
20.6% (18.4-22.9%)<br />
Unit-dose ward: 1.6% (1.0-2.5%)<br />
Multi-dose ward: 7.4% (6.1-8.9%)§<br />
Before vs. after implementation <strong>of</strong> unit-dose<br />
system: 37.2 vs. 38.5%;<br />
42.9 vs. 23.3%;<br />
20.1% vs. 7.8%;<br />
38.5% vs. 23.1%<br />
84 errors among 2756 observations in UK<br />
hospital using traditional ward stock system:<br />
3.0% (2.4-3.7%)<br />
63 errors among 919 observations in US hospital<br />
using unit-doses <strong>and</strong> automated dispensing: 6.9%<br />
(5.2-8.5%)<br />
Absolute difference: 3.9% (2.1-5.7%)<br />
UK hospital using traditional ward stock system:<br />
8.0% (6.2-9.8%)<br />
German hospital using traditional ward stock<br />
system: 5.1% (4.4-5.8)<br />
German hospital using unit-dose system: 2.4%<br />
(2.0-2.8%)<br />
Omission was the most common type <strong>of</strong> error<br />
* CI indicates confidence interval.<br />
† Errors <strong>of</strong> commission include administration <strong>of</strong> wrong dose or wrong or unordered drug, whereas errors<br />
<strong>of</strong> omission include missed doses for inclusion in a patient’s unit-dose drawer or a dose not<br />
administered.<br />
‡ As outlined in the text, the similarities in study setting, time, design <strong>and</strong> results suggest that these 2<br />
references contain data from the same study; information from these references was therefore combined<br />
<strong>and</strong> treated as a single study.<br />
§ The 95% CIs shown in the table were calculated using the reported data: 20 errors in 1234 observed<br />
doses on the unit-dose ward vs. 105 errors in 1428 observed doses on the multidose ward.
When wrong time errors were omitted, the above results changed so that the change to a unit-dose was<br />
associated with a significant increase in errors at the first hospital, a non-significant decrease in errors at<br />
the second hospital, <strong>and</strong> significant decreases in errors at the other two hospitals.<br />
107
References<br />
1. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events: Implications for prevention. JAMA.<br />
1995;274:29-34.<br />
2. Ringold DJ, Santell JP, Schneider PJ, Arenberg S. ASHP national survey <strong>of</strong> pharmacy<br />
practice in acute care settings: prescribing <strong>and</strong> transcribing–1998. American Society <strong>of</strong><br />
Health-System Pharmacists. Am J Health Syst Pharm. 1999;56:142-157.<br />
3. Facchinetti NJ, Campbell GM, Jones DP. Evaluating dispensing error detection rates in a<br />
hospital pharmacy. Med Care. 1999;37:39-43.<br />
4. Leape L, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems<br />
analysis <strong>of</strong> adverse drug events. JAMA. 1995;274:35-43.<br />
5. Tissot E, Cornette C, Demoly P, Jacquet M, Barale F, Capellier G. Medication errors at the<br />
administration stage in an intensive care unit. Intensive Care Med. 1999;25:353-359.<br />
6. Bates DW, Boyle DL, V<strong>and</strong>er Vliet MB, Schneider J, Leape L. Relationship between<br />
medication errors <strong>and</strong> adverse drug events. J Gen Intern Med. 1995;10:199-205.<br />
7. Santell JP. ASHP national survey <strong>of</strong> hospital-based pharmaceutical services–1994. Am J<br />
Health Syst Pharm. 1995;52:1179-1198.<br />
8. Godwin HN, Scott BE. Institutional patient care. In: Gennaro AR, editor. Remington: The<br />
Science <strong>and</strong> Practice <strong>of</strong> Pharmacy. Baltimore: Lippincott Williams & Wilkins; 2000:1911-<br />
31.<br />
9. Ringold DJ, Santell JP, Schneider PJ. ASHP national survey <strong>of</strong> pharmacy practice in acute<br />
care settings: dispensing <strong>and</strong> administration–1999. Am J Health Syst Pharm.<br />
2000;57:1759-1775.<br />
10. Read OK. Unit dose packaging <strong>of</strong> respiratory therapy solutions. Am J Health Syst Pharm.<br />
1975;32:49-51.<br />
11. Reitberg DP, Miller RJ, Bennes JF. Evaluation <strong>of</strong> two concurrent drug-delivery systems in<br />
a skilled-nursing facility. Am J Health Syst Pharm. 1982;39:1316-1320.<br />
12. Shultz SM, White SJ, Latiolais CJ. Medication errors reduced by unit-dose. Hospitals.<br />
1973;47:107-112.<br />
13. Means BJ, Derewicz HJ, Lamy PP. Medication errors in a multidose <strong>and</strong> computer-based<br />
unit dose drug distribution system. Am J Hosp Pharm. 1975;32:186-191.<br />
14. Simborg DW, Derewicz HJ. A highly automated hospital medication system: Five year's<br />
experience <strong>and</strong> evaluation. Ann Intern Med. 1975;83:342-346.<br />
15. A study <strong>of</strong> patient care involving a unit dose system. Final report: University <strong>of</strong> Iowa;<br />
1966. U.S.P.H.S. Grant # HM-00328-01.<br />
16. Greth PA, Tester WW, Black HJ. Decentralized pharmacy operations utilizing the unit<br />
dose concept. II. Drug information services <strong>and</strong> utilization in a decentralized pharmacy<br />
substation. Am J Hosp Pharm. 1965;22:558-563.<br />
17. Black HJ, Tester WW. Decentralized pharmacy operations utilizing the unit dose concept.<br />
3. Am J Hosp Pharm. 1967;24:120-129.<br />
18. Black HJ, Tester WW. The unit dose drug-distribution system. J Iowa Med Soc.<br />
1968;58:1174-1176.<br />
19. Slater WE, Hripko JR. The unit-dose system in a private hospital. 1. Implementation. Am J<br />
Hosp Pharm. 1968;25:408-417.<br />
20. Slater WE, Hripko JR. The unit-dose system in a private hospital. II. Evaluation. Am J<br />
Hosp Pharm. 1968;25:641-648.<br />
108
21. Barker KN, McConnell WE. The problems <strong>of</strong> detecting medication errors in hospitals. Am<br />
J Hosp Pharm. 1962;19:361-369.<br />
22. Dean BS, Allan EL, Barber ND, Barker KN. Comparison <strong>of</strong> medication errors in an<br />
American <strong>and</strong> British hospital. Am J Health Syst Pharm. 1995;52:2543-2549.<br />
23. Hynniman CE, Conrad WF, Urch WA, Rudnick BR, Parker PF. A comparison <strong>of</strong><br />
medication errors under the University <strong>of</strong> Kentucky unit dose system <strong>and</strong> traditional drug<br />
distribution systems in four hospitals. Am J Hosp Pharm. 1970;27:803-814.<br />
24. Schnell BR. A study <strong>of</strong> unit-dose drug distribution in four Can<strong>and</strong>ian hospitals. Can J<br />
Hosp Pharm. 1976;29:85-90.<br />
25. Taxis K, Dean B, Barber N. Hospital drug distribution systems in the UK <strong>and</strong> Germany - a<br />
study <strong>of</strong> medication errors. Pharm World Sci. 1998;21:25-31.<br />
109
110
Chapter 11. Automated Medication Dispensing Devices<br />
Michael D. Murray, PharmD, MPH<br />
Purdue University School <strong>of</strong> Pharmacy<br />
Background<br />
In the 1980s, automated dispensing devices appeared on the scene, a generation after the<br />
advent <strong>of</strong> unit-dose dispensing (Chapter 11). The invention <strong>and</strong> production <strong>of</strong> these devices<br />
brought hopes <strong>of</strong> reduced rates <strong>of</strong> medication errors, increased efficiency for pharmacy <strong>and</strong><br />
nursing staff, ready availability <strong>of</strong> medications where they are most <strong>of</strong>ten used (the nursing unit<br />
or inpatient ward), <strong>and</strong> improved pharmacy inventory <strong>and</strong> billing functions. 1-4 Although the<br />
capacity <strong>of</strong> such systems to contribute to patient safety appears great, surprisingly few studies<br />
have evaluated the clinical impact <strong>of</strong> these devices.<br />
Practice Description<br />
Automated dispensing systems are drug storage devices or cabinets that electronically<br />
dispense medications in a controlled fashion <strong>and</strong> track medication use. Their principal advantage<br />
lies in permitting nurses to obtain medications for inpatients at the point <strong>of</strong> use. Most systems<br />
require user identifiers <strong>and</strong> passwords, <strong>and</strong> internal electronic devices track nurses accessing the<br />
system, track the patients for whom medications are administered, <strong>and</strong> provide usage data to the<br />
hospital’s financial <strong>of</strong>fice for the patients’ bills.<br />
These automated dispensing systems can be stocked by centralized or decentralized<br />
pharmacies. Centralized pharmacies prepare <strong>and</strong> distribute medications from a central location<br />
within the hospital. Decentralized pharmacies reside on nursing units or wards, with a single<br />
decentralized pharmacy <strong>of</strong>ten serving several units or wards. These decentralized pharmacies<br />
usually receive their medication stock <strong>and</strong> supplies from the hospital’s central pharmacy.<br />
More advanced systems provide additional information support aimed at enhancing<br />
patient safety through integration into other external systems, databases, <strong>and</strong> the Internet. Some<br />
models use machine-readable code for medication dispensing <strong>and</strong> administration. Three types <strong>of</strong><br />
automated dispensing devices were analyzed in the studies reviewed here, the McLaughlin<br />
dispensing system, the Baxter ATC-212 dispensing system, <strong>and</strong> the Pyxis Medstation Rx. Their<br />
attributes are described below.<br />
• The McLaughlin dispensing system 5 includes a bedside dispenser, a<br />
programmable magnetic card, <strong>and</strong> a pharmacy computer. It is a locked system<br />
that is loaded with the medications prescribed for a patient. At the appropriate<br />
dosing time, the bedside dispenser drawer unlocks automatically to allow a<br />
dose to be removed <strong>and</strong> administered. A light above the patient’s door<br />
illuminates at the appropriate dosing time. Only certain medications fit in the<br />
compartmentalized cabinet (such as tablets, capsules, small pre-filled<br />
syringes, <strong>and</strong> ophthalmic drops).<br />
• The Baxter ATC-212 dispensing system 6 uses a microcomputer to pack unitdose<br />
tablets <strong>and</strong> capsules for oral administration. It is usually installed at the<br />
pharmacy. Medications are stored in calibrated canisters that are designed<br />
specifically for each medication. Canisters are assigned a numbered location,<br />
which is thought to reduce mix-up errors upon dispensing. When an order is<br />
sent to the microcomputer, a tablet is dispensed from a particular canister. The<br />
111
drug is ejected into a strip-packing device where it is labeled <strong>and</strong> hermetically<br />
sealed.<br />
• The Pyxis Medstation, Medstation Rx, <strong>and</strong> Medstation Rx 1000 are automated<br />
dispensing devices kept on the nursing unit. 7-9 These machines are <strong>of</strong>ten<br />
compared to automatic teller machines (ATMs). The Medstation interfaces<br />
with the pharmacy computer. Physicians’ orders are entered into the pharmacy<br />
computer <strong>and</strong> then transferred to the Medstation where patient pr<strong>of</strong>iles are<br />
displayed to the nurse who accesses the medications for verified orders. Each<br />
nurse is provided with a password that must be used to access the Medstation.<br />
Pharmacists or technicians keep these units loaded with medication. Charges<br />
are made automatically for drugs dispensed by the unit. Earlier models had<br />
sufficient memory to contain data for about one week, <strong>and</strong> newer models can<br />
store data for longer periods.<br />
Studies reviewed did not include the automated dispensing systems manufactured by<br />
Omnicell, which produces point-<strong>of</strong>-use systems that can be integrated into a hospital’s<br />
information system. 10 Omnicell systems are also capable <strong>of</strong> being integrated into external<br />
support systems that support machine-readable code, drug information services, <strong>and</strong> medication<br />
error reporting systems.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Medication errors within hospitals occur with 2% to 17% <strong>of</strong> doses ordered for<br />
inpatients. 5,7,11-14 It has been suggested that the rate <strong>of</strong> inpatient medication errors is one per<br />
patient per inpatient day. 15 The specific medication errors targeted by automated dispensing<br />
systems are those related to drug dispensing <strong>and</strong> administration. Even with the use <strong>of</strong> unit-doses<br />
(see Chapter 11) errors still occur at the dispensing 16 <strong>and</strong> administration stages 3,17 <strong>of</strong> the<br />
medication use process. For instance, in one large study <strong>of</strong> 530 medical errors in 10,070 written<br />
orders for drugs (5.3 errors/100 orders), 18 pharmacy dispensing accounted for 11% <strong>of</strong> errors <strong>and</strong><br />
nursing administration 38%. 3<br />
Opportunities for Impact<br />
Automated dispensing devices have become increasingly common either to supplement<br />
or replace unit-dose distribution systems in an attempt to improve medication availability,<br />
increase the efficiency <strong>of</strong> drug dispensing <strong>and</strong> billing, <strong>and</strong> reduce errors. A 1999 national survey<br />
<strong>of</strong> drug dispensing <strong>and</strong> administration practices indicated that 38% <strong>of</strong> responding hospitals used<br />
automated medication dispensing units <strong>and</strong> 8.2% used machine-readable coding with<br />
dispensing. 19 Three-fourths <strong>of</strong> respondents stated that their pharmacy was centralized <strong>and</strong> <strong>of</strong><br />
these centralized pharmacies, 77% were not automated. Hospitals with automated centralized<br />
pharmacies reported that greater than 50% <strong>of</strong> their inpatient doses were dispensed via centralized<br />
automated systems. Half <strong>of</strong> all responding hospitals used a decentralized medication storage<br />
system. One-third <strong>of</strong> hospitals with automated storage <strong>and</strong> dispensing systems were linked to the<br />
pharmacy computer. Importantly, about half <strong>of</strong> the surveyed hospitals reported drug distributions<br />
that bypassed the pharmacy including floor stock, borrowing patients’ medications, <strong>and</strong> hidden<br />
drug supplies.<br />
Study Designs<br />
There were no true r<strong>and</strong>omized trials. One crossover study <strong>of</strong> the McLaughlin dispensing<br />
system r<strong>and</strong>omized nurses to work with the intervention medication system or the control<br />
112
system. 5 We classified this as a Level 2 study, since, from the patient perspective, the design is<br />
that <strong>of</strong> a non-r<strong>and</strong>omized trial. Other studies included in this review consisted <strong>of</strong> retrospective<br />
observational studies with before-after 6-8 or cross-sectional design 11 (Level 3). The reviewed<br />
studies described dispensing systems for orally administered medications, <strong>and</strong> were published<br />
between 1984 <strong>and</strong> 1995 (see Table 12.1).<br />
Study Outcomes<br />
All studies measured rates <strong>of</strong> medication errors (Level 2 outcome). Four studies 5,7,8,11<br />
detected errors by direct observation using a methodology that was first described by Barker. 5<br />
Direct observation methods have been criticized because <strong>of</strong> purported Hawthorne effect (bias<br />
involving changed behavior resulting from measurements requiring direct observation <strong>of</strong> study<br />
subjects). However, proponents <strong>of</strong> the method state that such effects are short-lived, dissipating<br />
within hours <strong>of</strong> observation. 15 Dean <strong>and</strong> Barber have recently demonstrated the validity <strong>and</strong><br />
reliability <strong>of</strong> direct observational methods to detect medication administration errors. 20 Another<br />
study, a Level 3 design, determined errors by inspecting dispensed drugs. 6<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The evidence provided by the limited number <strong>of</strong> available, generally poor quality studies<br />
does not suggest that automated dispensing devices reduce medication errors. There is also no<br />
evidence to suggest that outcomes are improved with the use <strong>of</strong> these devices. Most <strong>of</strong> the<br />
published studies comparing automated devices with unit-dose dispensing systems report<br />
reductions in medication errors <strong>of</strong> omission <strong>and</strong> scheduling errors with the former. 7,9 The studies<br />
suffer from multiple problems with confounding, as they <strong>of</strong>ten compare hospitals or nursing care<br />
units that may differ in important respects other than the medication distribution system.<br />
Potential for Harm<br />
Human intervention may prevent these systems from functioning as designed.<br />
Pharmacists <strong>and</strong> nurses can override some <strong>of</strong> the patient safety features. When the turn around<br />
time for order entry into the automated system is prolonged, nurses may override the system<br />
thereby defeating its purpose. Furthermore, the automated dispensing systems must be refilled<br />
intermittently to replenish exhausted supplies. Errors can occur during the course <strong>of</strong> refilling<br />
these units or medications may shift from one drawer or compartment to another causing<br />
medication mix-ups. Either <strong>of</strong> these situations can slip past the nurse at medication<br />
administration.<br />
The results <strong>of</strong> the study <strong>of</strong> the McLaughlin dispensing system indicated that though<br />
overall errors were reduced compared to unit-dose (10.6% vs. 15.9%), errors decreased for 13 <strong>of</strong><br />
20 nurses but increased for the other 7 nurses. 5 In a study <strong>of</strong> Medstation Rx vs. unit-dose, 8 errors<br />
decreased in the cardiovascular surgery unit, where errors were recorded by work measurement<br />
observations. However, errors increased over 30% in 6 <strong>of</strong> 7 nurses after automated dispensing<br />
was installed in the cardiovascular intensive care unit, where incident reports <strong>and</strong> medication<br />
error reports were both used for ascertaining errors, raising the question <strong>of</strong> measurement bias.<br />
Finally, in a study primarily aimed at determining differences in errors for ward <strong>and</strong> unit-dose<br />
dispensing systems, 11 a greater error prevalence was found for medications dispensed using<br />
Medstation Rx compared with those dispensed using unit-dose or non-automated floor stock<br />
(17.1% vs. 5.4%).<br />
113
Costs <strong>and</strong> Implementation<br />
The cost <strong>of</strong> automated dispensing mainly involves the capital investment <strong>of</strong> renting or<br />
purchasing equipment for dispensing, labeling, <strong>and</strong> tracking (which <strong>of</strong>ten is done by computer).<br />
A 1995 study revealed that the cost <strong>of</strong> Medstation Rx to cover 10 acute care units (330 total<br />
beds) <strong>and</strong> 4 critical care units (48 total beds) in a large referral hospital would be $1.28 million<br />
over 5 years. Taking into account costs saved from reduced personnel <strong>and</strong> decreased drug waste,<br />
the units had the potential to save $1 million over 5 years. Most studies that examine economic<br />
impact found a trade-<strong>of</strong>f between reductions in medication dispensing time for pharmacy <strong>and</strong><br />
medication administration time for nursing personnel. A common complaint by nurses is long<br />
waiting lines at Pyxis Medstations if there are not enough machines. Nurses must access these<br />
machines using a nurse-specific password. This limited access to drugs on nursing units<br />
decreases drug waste <strong>and</strong> pilferage.<br />
Comment<br />
Although the implementation <strong>of</strong> automated dispensing reduces personnel time for<br />
medication administration <strong>and</strong> improves billing efficiency, reduction in medication errors have<br />
not been uniformly realized. Indeed, some studies suggest that errors may increase with some<br />
forms <strong>of</strong> automation. The results <strong>of</strong> the study <strong>of</strong> the McLaughlin Dispensing System by Barker et<br />
al 5 showed considerable nurse-to-nurse variability in the error rate between the automated<br />
system <strong>and</strong> conventional unit dose. Qualitative data aimed at determining the reason for this<br />
variability would be useful. The study by Klein et al 6 indicated little difference in the accuracy <strong>of</strong><br />
medication cart filling by the Baxter ATC-212 (0.65%) versus filling by technicians (0.84%).<br />
Borel <strong>and</strong> Rascati found that medication errors, largely those related to the time <strong>of</strong><br />
administration, were fewer after implementation <strong>of</strong> the Pyxis Medstation Rx (10.4%) compared<br />
with the historical period (16.9%). 7 These results are consistent with a more recent study by<br />
Shirley, that found a 31% increase in the on-time administration <strong>of</strong> scheduled doses after<br />
installation <strong>of</strong> the Medstation Rx 1000. 9 In contrast, errors were greater after Medstation Rx in<br />
the study by Schwarz <strong>and</strong> Brodowy, 8 increasing on 6 <strong>of</strong> 7 nursing units by more than 30%.<br />
Finally, Dean et al found half the errors in a ward-based system without automation in the<br />
United Kingdom (3.0%, 95% CI: 2.4-3.7%) compared with an automated unit-dose medication<br />
distribution system in the United States (6.9%, 95% CI: 5.2-8.5%). 11<br />
The practical limitations <strong>of</strong> the systems were illustrated by a variety <strong>of</strong> process deviations<br />
observed by Borel <strong>and</strong> Rascati. 7 These included nurses waiting at busy administration times,<br />
removal <strong>of</strong> doses ahead <strong>of</strong> time to circumvent waiting, <strong>and</strong> overriding the device when a dose<br />
was needed quickly. These procedural failures emphasize an <strong>of</strong>ten-raised point with the<br />
introduction <strong>of</strong> new technologies, namely that the latest innovations are not a solution for<br />
inadequate or faulty processes or procedures. 2<br />
Although automated dispensing systems are increasingly common, it appears they may<br />
not be completely beneficial in their current form. Further study is needed to demonstrate the<br />
effectiveness <strong>of</strong> newer systems such as the Omnicell automated dispensing devices. If the<br />
st<strong>and</strong>ard, namely unit-dose dispensing, is to be improved, such improvements will likely derive<br />
from robotics <strong>and</strong> informatics. To document impact <strong>of</strong> automated dispensing devices on patient<br />
safety, studies are needed comparing unit-dose dispensing with automated dispensing devices.<br />
Until the benefits <strong>of</strong> automated dispensing devices become clearer, the opportunities for impact<br />
<strong>of</strong> these devices is uncertain.<br />
114
Table 11.1. Six studies reviewing automated drug dispensing systems*<br />
Study Study Design Study Outcomes N Results<br />
Barker,<br />
1984 5<br />
Klein,<br />
1994 6<br />
Borel,<br />
1995 7<br />
Schwarz,<br />
1995 8<br />
Dean,<br />
1995 11<br />
Prospective<br />
controlled<br />
clinical trial<br />
(Level 2)<br />
Prospective<br />
comparison <strong>of</strong><br />
two cohorts<br />
(Level 2)<br />
Prospective<br />
before-after study<br />
(Level 2)<br />
Prospective<br />
before-after study<br />
(Level 2)<br />
Cross-sectional<br />
comparison<br />
(Level 3) <strong>of</strong> US<br />
<strong>and</strong> UK hospitals<br />
with different<br />
pharmacy<br />
distribution<br />
systems<br />
* CI indicates confidence interval.<br />
† Study used various denominator data.<br />
Errors <strong>of</strong> omission <strong>and</strong><br />
commission among<br />
number <strong>of</strong> ordered <strong>and</strong><br />
unauthorized doses.<br />
(Level 2)<br />
Dispensing errors in<br />
unit-dose drawers to be<br />
delivered to nursing<br />
units (Level 2)<br />
Errors observed during<br />
medication<br />
administrationin<br />
medications<br />
administered (Level 2)<br />
Errors in medications<br />
administered (Level 2)<br />
Errors in medications<br />
administered (Level 2)<br />
115<br />
1775 96 errors among 902 observations<br />
(10.6%) using the McLaughlin<br />
dispensing system vs. 139 errors<br />
among 873 observations (15.9%)<br />
using unit-dose dispensing (control)<br />
7842 34 errors found among 4029 doses<br />
(0.84%) filled manually by<br />
technicians vs. 25 errors among 3813<br />
doses (0.66%) filled by automated<br />
dispensing device<br />
1802 148 errors among 873 observations<br />
(16.9%) before vs. 97 errors among<br />
929 observations (10.4%) after<br />
Medstation Rx (p
References<br />
1. Perini VJ, Vermeulen LC, Jr. Comparison <strong>of</strong> automated medication-management systems.<br />
Am J Hosp Pharm. 1994;51:1883-1891.<br />
2. Murray MD. Information technology: the infrastructure for improvements to the<br />
medication-use process. Am J Health Syst Pharm. 2000;57:565-571.<br />
3. Guerrero RM, Nickman NA, Jorgenson JA. Work activities before <strong>and</strong> after<br />
implementation <strong>of</strong> an automated dispensing system. Am J Health Syst Pharm.<br />
1996;53:548-554.<br />
4. Felkey BG, Barker KN. Technology <strong>and</strong> automation in pharmaceutical care. J Am Pharm<br />
Assoc (Wash). 1996;NS36:309-314.<br />
5. Barker KN, Pearson RE, Hepler CD, Smith WE, Pappas CA. Effect <strong>of</strong> an automated<br />
bedside dispensing machine on medication errors. Am J Hosp Pharm. 1984;41:1352-1358.<br />
6. Klein EG, Santora JA, Pascale PM, Kitrenos JG. Medication cart-filling time, accuracy,<br />
<strong>and</strong> cost with an automated dispensing system. Am J Hosp Pharm. 1994;51:1193-1196.<br />
7. Borel JM, Rascati KL. Effect <strong>of</strong> an automated, nursing unit-based drug-dispensing device<br />
on medication errors . Am J Health Syst Pharm. 1995;52:1875-1879.<br />
8. Schwarz HO, Brodowy BA. Implementation <strong>and</strong> evaluation <strong>of</strong> an automated dispensing<br />
system. Am J Health Syst Pharm. 1995;52:823-828.<br />
9. Shirley KL. Effect <strong>of</strong> an automated dispensing system on medication administration time.<br />
Am J Health Syst Pharm. 1999;56:1542-1545.<br />
10. Omnicell Automated Systems. Omnicell Homepage. Available at:<br />
http://www.omnicell.com. Accessed May 28, 2001.<br />
11. Dean BS, Allan EL, Barber ND, Barker KN. Comparison <strong>of</strong> medication errors in an<br />
American <strong>and</strong> British hospital. Am J Health Syst Pharm. 1995;52:2543-2549.<br />
12. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence <strong>of</strong> adverse<br />
drug events <strong>and</strong> potential adverse drug events: Implications for prevention. JAMA.<br />
1995;274:29-34.<br />
13. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in<br />
hospitalized patients. Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA.<br />
1997;277:301-306.<br />
14. Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable<br />
adverse drug events in hospitalized patients: a comparative study <strong>of</strong> intensive care <strong>and</strong><br />
general care units. Crit Care Med. 1997;25:1289-1297.<br />
15. Allan EL, Barker KN. Fundamentals <strong>of</strong> medication error research. Am J Hosp Pharm.<br />
1990;47:555-571.<br />
16. Facchinetti NJ, Campbell GM, Jones DP. Evaluating dispensing error detection rates in a<br />
hospital pharmacy. Med Care. 1999;37:39-43.<br />
17. Tissot E, Cornette C, Demoly P, Jacquet M, Barale F, Capellier G. Medication errors at the<br />
administration stage in an intensive care unit. Intensive Care Med. 1999;25:353-3359.<br />
18. Bates DW, Boyle DL, V<strong>and</strong>er Vliet MB, Schneider J, Leape L. Relationship between<br />
medication errors <strong>and</strong> adverse drug events. J Gen Intern Med. 1995;10:199-205.<br />
19. Ringold DJ, Santell JP, Schneider PJ. ASHP national survey <strong>of</strong> pharmacy practice in acute<br />
care settings: dispensing <strong>and</strong> administration–1999. Am J Health Syst Pharm.<br />
2000;57:1759-1775.<br />
20. Dean B, Barber N. Validity <strong>and</strong> reliability <strong>of</strong> observational methods for studying<br />
medication administration errors. Am J Health Syst Pharm. 2001;58:54-59.<br />
116
117
Section B. Infection Control<br />
Chapter 12. <strong>Practices</strong> to Improve H<strong>and</strong>washing Compliance<br />
Chapter 13. Impact <strong>of</strong> Barrier Precautions in Reducing the Transmission <strong>of</strong> Serious<br />
Nosocomial Infections<br />
Chapter 14. Impact <strong>of</strong> Changes in Antibiotic Use <strong>Practices</strong> on Nosocomial Infections <strong>and</strong><br />
Antimicrobial Resistance – Clostridium Difficile <strong>and</strong> Vancomycin-resistant<br />
Enterococcus (VRE)<br />
Chapter 15. Prevention <strong>of</strong> Nosocomial Urinary Tract Infections<br />
Subchapter 15.1. Use <strong>of</strong> Silver Alloy Urinary Catheters<br />
Subchapter 15.2. Use <strong>of</strong> Suprapubic Catheters<br />
Chapter 16. Prevention <strong>of</strong> Intravascular Catheter-Associated Infections<br />
Subchapter 16.1. Use <strong>of</strong> Maximum Barrier Precautions during Central Venous<br />
Catheter Insertion<br />
Subchapter 16.2. Use <strong>of</strong> Central Venous Catheters Coated with Antibacterial or<br />
Antiseptic Agents<br />
Subchapter 16.3. Use <strong>of</strong> Chlorhexidine Gluconate at the Central Venous<br />
Catheter Insertion Site<br />
Subchapter 16.4. Other <strong>Practices</strong><br />
Chapter 17. Prevention <strong>of</strong> Ventilator-Associated Pneumonia (VAP)<br />
Subchapter 17.1. <strong>Patient</strong> Positioning: Semi-recumbent Positioning <strong>and</strong><br />
Continuous Oscillation<br />
Subchapter 17.2. Continuous Aspiration <strong>of</strong> Subglottic Secretions<br />
Subchapter 17.3. Selective Digestive Tract Decontamination<br />
Subchapter 17.4. Sucralfate <strong>and</strong> Prevention <strong>of</strong> VAP<br />
118
Chapter 12. <strong>Practices</strong> to Improve H<strong>and</strong>washing Compliance<br />
Ebbing Lautenbach, MD, MPH, MSCE<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Background<br />
Hospital-acquired infections exact a tremendous toll, resulting in increased morbidity <strong>and</strong><br />
mortality, <strong>and</strong> increased health care costs. 1,2 Since most hospital-acquired pathogens are<br />
transmitted from patient to patient via the h<strong>and</strong>s <strong>of</strong> health care workers, 3 h<strong>and</strong>washing is the<br />
simplest <strong>and</strong> most effective, proven method to reduce the incidence <strong>of</strong> nosocomial infections. 4<br />
Indeed, over 150 years ago, Ignaz Semmelweis demonstrated that infection-related mortality<br />
could be reduced when health care personnel washed their h<strong>and</strong>s. 5 A recent review summarized<br />
the 7 studies published between 1977 <strong>and</strong> 1995 that examined the relationship between h<strong>and</strong><br />
hygiene <strong>and</strong> nosocomial infections. 6<br />
Most <strong>of</strong> the reports analyzed in this study reveal a temporal relation between improved<br />
h<strong>and</strong> hygiene <strong>and</strong> reduced infection rates. 6 Despite this well-established relationship, compliance<br />
with h<strong>and</strong>washing among all types <strong>of</strong> health care workers remains poor. 7-11 Identifying effective<br />
methods to improve the practice <strong>of</strong> h<strong>and</strong>washing would greatly enhance the care <strong>of</strong> patients <strong>and</strong><br />
result in a significant decrease in hospital-acquired infections.<br />
Practice Description<br />
This chapter focuses on practices that increase compliance with h<strong>and</strong>washing, rather than<br />
the already proven efficacy <strong>of</strong> h<strong>and</strong>washing itself. 4 The term “h<strong>and</strong>washing” defines several<br />
actions designed to decrease h<strong>and</strong> colonization with transient microbiological flora, achieved<br />
either through st<strong>and</strong>ard h<strong>and</strong>washing or h<strong>and</strong> disinfection. 4 St<strong>and</strong>ard h<strong>and</strong>washing refers to the<br />
action <strong>of</strong> washing h<strong>and</strong>s in water with detergent to remove dirt <strong>and</strong> loose, transient flora. H<strong>and</strong><br />
disinfection refers to any action where an antiseptic solution is used to clean the h<strong>and</strong>s (ie,<br />
medicated soap or alcohol). H<strong>and</strong>washing with bl<strong>and</strong> soap (without disinfectant) is inferior to<br />
h<strong>and</strong>washing with a disinfecting agent. 12 Hygienic h<strong>and</strong> rub consists <strong>of</strong> rubbing h<strong>and</strong>s with a<br />
small quantity (2-3mL) <strong>of</strong> a highly effective <strong>and</strong> fast acting antiseptic agent. Because alcohols<br />
have excellent antimicrobial properties <strong>and</strong> the most rapid action <strong>of</strong> all antiseptics, they are the<br />
preferred agents for hygienic h<strong>and</strong> rub (also called waterless h<strong>and</strong> disinfection). Also, alcohols<br />
dry very rapidly, allowing for faster h<strong>and</strong> disinfection. 4<br />
Given health care workers’ documented low compliance with recommended<br />
h<strong>and</strong>washing practices, 7-9 improving compliance represents a more pressing patient safety<br />
concern than does the choice <strong>of</strong> different disinfectants, or attention to other specific issues such<br />
as choice <strong>of</strong> drying method, removal <strong>of</strong> rings, etc. Of the 14 studies reviewed in this chapter<br />
(Table 12.1), all study sites utilized hygienic h<strong>and</strong> rub <strong>and</strong>/or another method <strong>of</strong> h<strong>and</strong><br />
disinfection as st<strong>and</strong>ard practice. However, only 2 studies assessed the specific characteristics <strong>of</strong><br />
h<strong>and</strong>washing practice (eg, duration <strong>of</strong> washing, method <strong>of</strong> drying) according to established<br />
hospital guidelines, 13,14 while the other 12 studies assessed only whether or not h<strong>and</strong>washing<br />
occurred after patient contact.<br />
119
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Nosocomial infections occur in about 7-10% <strong>of</strong> hospitalized patients 1 <strong>and</strong> account for<br />
approximately 80,000 deaths per year in the United States. 15 Although h<strong>and</strong>washing has been<br />
proven to be the single most effective method to reduce nosocomial infections, compliance with<br />
recommended h<strong>and</strong> hygiene practices is unacceptably low. 7-9 Indeed, a recent review <strong>of</strong> 11<br />
studies noted that the level <strong>of</strong> compliance with basic h<strong>and</strong>washing ranged from 16% to 81%. 4 Of<br />
these 11 studies, only 2 noted compliance levels above 50%. 4 One reason for poor h<strong>and</strong>washing<br />
compliance may be that the importance <strong>of</strong> this simple protocol for decreasing infections is<br />
routinely underestimated by health care workers. 2 Recent surveys demonstrate that although<br />
most health care workers recognize the importance <strong>of</strong> h<strong>and</strong>washing in reducing infections, they<br />
routinely overestimate their own compliance with this procedure. 10 A survey <strong>of</strong> approximately<br />
200 health care workers noted that 89% recognized h<strong>and</strong>washing as an important means <strong>of</strong><br />
preventing infection. 10 Furthermore, 64% believed they washed their h<strong>and</strong>s as <strong>of</strong>ten as their<br />
peers, <strong>and</strong> only 2% believed that they washed less <strong>of</strong>ten than their peers did. 10<br />
Opportunities for Impact<br />
Given these findings, opportunities for improvement in current practice are substantial,<br />
<strong>and</strong> efforts to improve current practice would have vast applicability. Many risk factors for noncompliance<br />
with h<strong>and</strong> hygiene guidelines have been identified, including pr<strong>of</strong>essional category<br />
(eg, physician, nurse, technician), hospital ward, time <strong>of</strong> day or week, <strong>and</strong> type <strong>and</strong> intensity <strong>of</strong><br />
patient care. 8 These results suggest that interventions could be particularly targeted to certain<br />
groups <strong>of</strong> health care workers or to particular locations, to increase the likelihood <strong>of</strong> compliance.<br />
Importantly, this study demonstrates that the individuals with the highest need for h<strong>and</strong> hygiene<br />
(ie, those with the greatest workloads) were precisely the same group least likely to wash their<br />
h<strong>and</strong>s. Finally, another recent study noted that approximately 75% <strong>of</strong> health care workers<br />
surveyed reported that rewards or punishments would not improve h<strong>and</strong>washing, but 80%<br />
reported that easy access to sinks <strong>and</strong> availability <strong>of</strong> h<strong>and</strong> washing facilities would lead to<br />
increased compliance. 10<br />
Study Designs<br />
A structured search <strong>of</strong> the PubMed database (including MEDLINE) <strong>and</strong> review <strong>of</strong> the<br />
bibliographies <strong>of</strong> relevant articles identified 14 studies that have examined methods to improve<br />
h<strong>and</strong>washing compliance (Table 12.1). Three studies were non-r<strong>and</strong>omized controlled trials<br />
(Level 2) that directly compared separate units, or parts <strong>of</strong> units, in which one area received the<br />
intervention <strong>and</strong> another did not. 14,16,17 Eleven studies were before-after studies (Level 3), in<br />
which baseline data regarding h<strong>and</strong>washing rates were obtained during an initial observation<br />
period, <strong>and</strong> then measured again in the time period after a particular intervention. Regardless <strong>of</strong><br />
the type <strong>of</strong> study design, details regarding the comparability <strong>of</strong> the groups under observation<br />
were reported in only 4 studies. 13,16,18,19<br />
Study Outcomes<br />
All <strong>of</strong> the studies reported changes in percent compliance with h<strong>and</strong>washing, assessing<br />
whether or not h<strong>and</strong>washing took place (Table 12.1). While 13 studies assessed h<strong>and</strong>washing<br />
through observation <strong>of</strong> health care worker behavior (Level 2), one study assessed soap usage as<br />
an indicator <strong>of</strong> h<strong>and</strong>washing frequency (Level 3). 20 Two studies also assessed changes in the<br />
quality <strong>of</strong> h<strong>and</strong>washing. 13,14 Several studies reported results <strong>of</strong> surveys conducted following<br />
120
interventions to assess effectiveness <strong>and</strong> potential adverse events related to the<br />
interventions. 14,20,21 One study also assessed changes in 2 clinical outcomes (incidence <strong>of</strong><br />
nosocomial infections <strong>and</strong> newly detected cases <strong>of</strong> methicillin-resistant Staphylococcus aureus)<br />
as a result <strong>of</strong> interventions (Level 1). 18<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Since many different risk factors have been identified for non-compliance with<br />
h<strong>and</strong>washing, it is not surprising that a variety <strong>of</strong> different interventions have been studied in an<br />
effort to improve this practice. While most <strong>of</strong> the reviewed studies demonstrated significant<br />
improvement in h<strong>and</strong>washing compliance, 9,13,17,18,20-22 some did not. 14,19,23,24 No single strategy<br />
has consistently been shown to sustain improved compliance with h<strong>and</strong>washing protocols. 11 In<br />
fact, <strong>of</strong> the studies which assessed longer-term results following intervention, 16,21,25 all 3 found<br />
that compliance rates decreased from those immediately following the intervention, <strong>of</strong>ten<br />
approaching pre-intervention levels.<br />
Potential for Harm<br />
While no harm is likely to befall a patient as a result <strong>of</strong> h<strong>and</strong>washing, one potential<br />
adverse effect <strong>of</strong> h<strong>and</strong>washing for health care workers is skin irritation. Indeed, skin irritation<br />
constitutes an important barrier to appropriate compliance with h<strong>and</strong>washing guidelines. 27 Soaps<br />
<strong>and</strong> detergents can damage the skin when applied on a regular basis. Alcohol-based preparations<br />
are less irritating to the skin, <strong>and</strong> with the addition <strong>of</strong> emollients, may be tolerated better. 6<br />
Another potential harm <strong>of</strong> increasing compliance with h<strong>and</strong>washing is the amount <strong>of</strong> time<br />
required to do it adequately. Current recommendations for st<strong>and</strong>ard h<strong>and</strong>washing suggest 15-30<br />
seconds <strong>of</strong> h<strong>and</strong>washing is necessary for adequate h<strong>and</strong> hygiene. 28 Given the many times during<br />
a nursing shift that h<strong>and</strong>washing should occur, this is a significant time commitment that could<br />
potentially impede the performance <strong>of</strong> other patient care duties. In fact, lack <strong>of</strong> time is one <strong>of</strong> the<br />
most common reasons cited for failure to wash h<strong>and</strong>s. 11 Since alcohol-based h<strong>and</strong>rubs require<br />
much less time, it has been suggested that they might resolve this concern. In fact, a recent study<br />
which modeled compliance time for h<strong>and</strong>washing as compared with alcoholic rubs, suggested<br />
that, given 100% compliance, h<strong>and</strong>washing would consume 16 hours <strong>of</strong> nursing time per<br />
st<strong>and</strong>ard day shift, while alcohol rub would consume only 3 hours. 29<br />
Costs <strong>and</strong> Implementation<br />
Interventions designed to improve h<strong>and</strong>washing may require significant financial <strong>and</strong><br />
human resources. This is true both for multifaceted educational/feedback initiatives, as well as<br />
for interventions that require capital investments in equipment such as more sinks, automated<br />
sinks, or new types <strong>of</strong> h<strong>and</strong> hygiene products. The costs incurred by such interventions must be<br />
balanced against the potential gain derived from reduced numbers <strong>of</strong> nosocomial infections.<br />
Only one study addressed the cost implications <strong>of</strong> h<strong>and</strong>washing initiatives. 20 The implementation<br />
<strong>of</strong> a patient education campaign, when compared to the estimated $5000 per episode cost <strong>of</strong> each<br />
nosocomial infection, would result in an annual savings <strong>of</strong> approximately $57,600 for a 300-bed<br />
hospital with 10,000 admissions annually. 20 As others have estimated that the attributable cost <strong>of</strong><br />
a single nosocomial bloodstream infection is approximately $40,000 per survivor, 30 the potential<br />
cost savings <strong>of</strong> interventions to improve h<strong>and</strong>washing may be even greater.<br />
121
Comment<br />
While many studies have investigated a variety <strong>of</strong> interventions designed to improve<br />
compliance with h<strong>and</strong>washing, the results have been mixed. Even when initial improvements in<br />
compliance have been promising, long-term continued compliance has been disappointing.<br />
Future studies should focus on more clearly identifying risk factors for non-compliance, <strong>and</strong><br />
designing interventions geared toward sustainability. Some investigators postulate that better<br />
underst<strong>and</strong>ing <strong>of</strong> behavior theory, <strong>and</strong> its application to infection control practices, might result<br />
in more effectively designed interventions. 26 In addition, any intervention must target reasons for<br />
non-compliance at all levels <strong>of</strong> health care (ie, individual, group, institution) in order to be<br />
effective. A more detailed study <strong>of</strong> the cost (<strong>and</strong> potentially cost savings) <strong>of</strong> h<strong>and</strong>washing<br />
initiatives would also foster greater enthusiasm among health care institutions to support such<br />
initiatives.<br />
122
Table 12.1. Fourteen studies <strong>of</strong> practices to improve h<strong>and</strong>washing compliance*<br />
Study Setting; Practice Study Design,<br />
Outcomes<br />
All medical staff in a neurologic ICU <strong>and</strong> a surgical<br />
ICU in a 350-bed tertiary care teaching hospital in<br />
Washington, DC, 1983-84; multifaceted intervention<br />
(education, automatic sinks, feedback) 16<br />
<strong>Medical</strong> staff in 2 ICUs in a university teach hospital in<br />
Philadelphia; increase number <strong>of</strong> available sinks 17<br />
<strong>Medical</strong> staff in a 6-bed post-anesthesia recovery room<br />
<strong>and</strong> a 15-bed neonatal ICU in a tertiary care hospital in<br />
Baltimore, 1990; automatic sink compared with<br />
st<strong>and</strong>ard sink 14<br />
All staff at a large acute-care teaching hospital in<br />
France, 1994-97; h<strong>and</strong> hygiene campaign including<br />
posters, feedback, <strong>and</strong> introduction <strong>of</strong> alcohol-based<br />
solution 18<br />
<strong>Medical</strong> staff in a 6-bed pediatric ICU in a large<br />
academic medical center in Virginia, 1982-83;<br />
m<strong>and</strong>atory gowning 19<br />
<strong>Medical</strong> staff in 2 ICUs in a community teaching<br />
hospital in Tennessee, 1983-84; sequential<br />
interventions <strong>of</strong> lectures, buttons, observation, <strong>and</strong><br />
feedback 24<br />
<strong>Medical</strong> staff in an 18-bed ICU in a tertiary care<br />
hospital in Australia; introduction <strong>of</strong> chlorhexidinebased<br />
antiseptic h<strong>and</strong>rub lotion 9<br />
12 nurses in a 12-bed ICU in Mississippi, 1990;<br />
education/feedback intervention 31<br />
<strong>Medical</strong> staff in an 18-bed pediatric ICU in a children’s<br />
teaching hospital in Melbourne, 1994; 5-step behavioral<br />
modification program 25<br />
<strong>Medical</strong> staff in a 3000-bed tertiary care center in<br />
France, 1994-95; 13-step h<strong>and</strong>washing protocol 13<br />
123<br />
Level 2, Level<br />
2<br />
Level 2, Level<br />
2<br />
Level 2, Level<br />
2<br />
Level 3, Level<br />
1<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
H<strong>and</strong>washing Compliance<br />
(unless otherwise noted)†<br />
69% vs. 59% (p=0.005)<br />
76% vs. 51% (p
Table 12.1. Fourteen studies <strong>of</strong> practices to improve h<strong>and</strong>washing compliance (cont.)<br />
<strong>Medical</strong> staff in two ICUs at a teaching hospital in<br />
Virginia, 1997; 6 education/feedback sessions followed<br />
by introduction <strong>of</strong> alcohol antiseptic agent 22<br />
<strong>Medical</strong> staff in a 14-bed ICU in a tertiary care<br />
hospital in France, 1998; introduction <strong>of</strong> alcohol-based<br />
solution 21<br />
All staff in a medical ICU <strong>and</strong> step-down unit in a large<br />
teaching hospital in Virginia; installation <strong>of</strong> alcoholbased<br />
solution 23<br />
<strong>Medical</strong> staff on 2 general inpatient floor at each <strong>of</strong> 4<br />
community hospitals in New Jersey; patient education<br />
intervention 20<br />
* ICU indicates intensive care unit.<br />
† Results are reported as intervention group vs. control group.<br />
124<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
2<br />
Level 3, Level<br />
3<br />
Baseline 22%;<br />
Education/feedback 25%;<br />
Alcohol antiseptic 48%;<br />
(p
References<br />
1. Haley RW, Culver DH, White JW, Morgan WM, Emori TG, Munn VP. The efficacy <strong>of</strong><br />
infection surveillance <strong>and</strong> control programs in preventing nosocomial infection in US<br />
hospitals. Am J Epidemiol. 1985;121:182-205.<br />
2. Jarvis WR. H<strong>and</strong>washing-the Semmelweis lesson forgotten? Lancet. 1994;344:1311-2.<br />
3. Larson E. A causal link between h<strong>and</strong>washing <strong>and</strong> risk <strong>of</strong> infection? Examination <strong>of</strong> the<br />
evidence. Infect Control Hosp Epidemiol. 1988;9:28-36.<br />
4. Pittet D. Improving compliance with h<strong>and</strong> hygiene in hospitals. Infect Control Hosp<br />
Epidemiol. 2000;21:381-386.<br />
5. Newsom SW. Pioneers in infection control. Ignaz Philipp Semmelweis. J Hosp Infect.<br />
1993;23:175-187.<br />
6. Larson E. Skin hygiene <strong>and</strong> infection prevention: more <strong>of</strong> the same or different<br />
approaches? Clin Infect Dis. 1999;29:1287-1294.<br />
7. Doebbeling BN, Stanley GL, Sheetz CT, Pfaller MA, Houston AK, Annis L, et al.<br />
Comparative efficacy <strong>of</strong> alternative h<strong>and</strong>washing agents in reducing nosocomial infections<br />
in intensive care units. N Engl J Med. 1992;327:88-93.<br />
8. Pittet D, Mourouga P, Perneger TV. Compliance with h<strong>and</strong>washing in a teaching hospital.<br />
Ann Intern Med. 1999;130:126-130.<br />
9. Graham M. Frequency <strong>and</strong> duration <strong>of</strong> h<strong>and</strong>washing in an intensive care unit. Am J Infect<br />
Control. 1990;18:77-80.<br />
10. Harris AD, Samore MH, Nafziger R, Dirosario K, Roghmann MC, Carmeli Y. A survey on<br />
h<strong>and</strong>washing practices <strong>and</strong> opinions <strong>of</strong> healthcare workers. J Hosp Infect. 2000;45:318-<br />
321.<br />
11. Larson E, Kretzer EK. Compliance with h<strong>and</strong>washing <strong>and</strong> barrier precautions. J Hosp<br />
Infect. 1995;30(Suppl):88-106.<br />
12. Ehrenkranz NJ, Alfonso BC. Failure <strong>of</strong> bl<strong>and</strong> soap h<strong>and</strong>wash to prevent h<strong>and</strong> transfer <strong>of</strong><br />
patient bacteria to urethral catheters. Infect Control Hosp Epidemiol. 1991;12:654-662.<br />
13. Coignard B, Gr<strong>and</strong>bastien B, Berrouane Y, Krembel C, Queverue M, Salomez JL, et al.<br />
H<strong>and</strong>washing quality: impact <strong>of</strong> a special program. Infect Control Hosp Epidemiol.<br />
1998;19:510-513.<br />
14. Larson E, McGeer A, Quraishi A, Krenzischek D, Parsons BJ, Holdford J, et al. Effect <strong>of</strong><br />
an automated sink on h<strong>and</strong>washing practices <strong>and</strong> attitudes in high-risk units. Infect Control<br />
Hosp Epidemiol. 1991;12:422-428.<br />
15. Jarvis WR. Selected aspects <strong>of</strong> the socioeconomic impact <strong>of</strong> nosocomial infections:<br />
morbidity, mortality, cost, <strong>and</strong> prevention. Infect Control Hosp Epidemiol. 1996;17:552-<br />
557.<br />
16. Larson EL, Bryan JL, Adler LM, Blane C. A multifaceted approach to changing<br />
h<strong>and</strong>washing behavior. Am J Infect Control. 1997;25:3-10.<br />
17. Kaplan LM, McGuckin M. Increasing h<strong>and</strong>washing compliance with more accessible<br />
sinks. Infect Control Hosp Epidemiol. 1986;7:408-410.<br />
18. Pittet D, Hugonnet S, Harbarth S, Mourouga P, Sauvan V, Touveneau S, et al.<br />
Effectiveness <strong>of</strong> a hospital-wide programme to improve compliance with h<strong>and</strong> hygiene.<br />
Lancet. 2000;356:1307-1312.<br />
19. Donowitz LG. H<strong>and</strong>washing technique in a pediatric intensive care unit. Am J Dis Child.<br />
1987;141:683-685.<br />
125
20. McGuckin M, Waterman R, Porten L, Bello S, Caruso M, Juzaitis B, et al. <strong>Patient</strong><br />
education model for increasing h<strong>and</strong>washing compliance. Am J Infect Control.<br />
1999;27:309-314.<br />
21. Maury E, Alzieu M, Baudel JL, Haram N, Barbut F, Guidet B, et al. Availability <strong>of</strong> an<br />
alcohol solution can improve h<strong>and</strong> disinfection compliance in an intensive care unit. Am J<br />
Respir Crit Care Med. 2000;162:324-327.<br />
22. Bisch<strong>of</strong>f WE, Reynolds TM, Sessler CN, Edmond MB, Wenzel RP. H<strong>and</strong>washing<br />
compliance by health care workers: the impact <strong>of</strong> introducing an accessible, alcohol-based<br />
h<strong>and</strong> antiseptic. Arch Intern Med. 2000;160:1017-1021.<br />
23. Muto CA, Sistrom MG, Farr BM. H<strong>and</strong> hygiene rates unaffected by installation <strong>of</strong><br />
dispensers <strong>of</strong> a rapidly acting h<strong>and</strong> antiseptic. Am J Infect Control. 2000;28:273-276.<br />
24. Simmons B, Bryant J, Km N, Spencer L, Arheart K. The role <strong>of</strong> h<strong>and</strong>washing in prevention<br />
<strong>of</strong> endemic intensive care unit infections. Infect Control Hosp Epidemiol. 1990;11:589-<br />
594.<br />
25. Tibballs J. Teaching hospital medical staff to h<strong>and</strong>wash. Med J Aust. 1996; 164: 395-398.<br />
26. Kretzer EK, Larson EL. Behavioral interventions to improve infection control practices.<br />
Am J Infect Control. 1998;26:245-253.<br />
27. Larson E. H<strong>and</strong>washing <strong>and</strong> skin: physiologic <strong>and</strong> bacteriologic aspects. Infect Control.<br />
1985;6:14-23.<br />
28. Larson EL. APIC Guidelines for h<strong>and</strong>washing <strong>and</strong> h<strong>and</strong> antisepsis in health care settings.<br />
Am J Infect Control. 1995;23:251-269.<br />
29. Voss A, Widmer AF. No time for h<strong>and</strong>washing!? H<strong>and</strong>washing versus alcoholic rub: can<br />
we afford 100% compliance? Infect Control Hosp Epidemiol. 1997;18:205-208.<br />
30. Pittet D, Tarara D, Wenzel RP. Nosocomial bloodstream infection in critically ill patients.<br />
Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA. 1994;271:1598-1601.<br />
31. Dubbert PM, Dolce J, Richter W, Miller M, Chapman SW. Increasing ICU staff<br />
h<strong>and</strong>washing: effects <strong>of</strong> education <strong>and</strong> group feedback. Infect Control Hosp Epidemiol.<br />
1990;11:191-193.<br />
126
Chapter 13. Impact <strong>of</strong> Barrier Precautions in Reducing the Transmission <strong>of</strong><br />
Serious Nosocomial Infections<br />
Ebbing Lautenbach, MD, MPH, MSCE<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Background<br />
Many nosocomial infections are easily transferable from patient-to-patient, either via the<br />
h<strong>and</strong>s <strong>of</strong> health care workers, 1,2 or through the contamination <strong>of</strong> inanimate objects, including<br />
clothing <strong>and</strong> equipment. 3,4 For some infections, the threat to other patients is considered serious<br />
enough that many institutions employ special barrier precautions, such as the use <strong>of</strong> gloves,<br />
gowns <strong>and</strong> disposable equipment for all patient contact, in caring for patients colonized or<br />
infected with these pathogens. Vancomycin-resistant enterococci (VRE) 5 <strong>and</strong> Clostridium<br />
difficile 6 represent 2 typical examples <strong>of</strong> nosocomial pathogens that may trigger such<br />
precautions.<br />
Although adherence to barrier precautions to prevent the spread <strong>of</strong> particularly<br />
concerning nosocomial pathogens has obvious face validity, the utility <strong>of</strong> specific interventions<br />
<strong>and</strong> the optimal forms they should take remain unclear. This uncertainty may in part reflect the<br />
impact <strong>of</strong> particular aspects <strong>of</strong> the epidemiology <strong>of</strong> the targeted nosocomial pathogens – ie, the<br />
benefit <strong>of</strong> a given strategy may vary in different settings <strong>and</strong> with different organisms.<br />
Consequently, this chapter contrasts with the review <strong>of</strong> h<strong>and</strong>washing (Chapter 13), a practice for<br />
which the benefit was regarded as sufficiently established to warrant focusing on strategies for<br />
improving compliance. While compliance with barrier precautions is also an important topic <strong>and</strong><br />
likely plays a significant role in the efficacy <strong>of</strong> such interventions, this chapter analyzes the<br />
literature evaluating the benefit <strong>of</strong> the barrier precautions themselves.<br />
Practice Description<br />
Barrier precautions include any activity designed to prevent the spread <strong>of</strong> nosocomial<br />
pathogens from patient to patient. This chapter reviews the following 3 practices:<br />
• Use <strong>of</strong> gowns <strong>and</strong> gloves for all contact with patients colonized or infected<br />
with VRE <strong>and</strong>/or C. difficile: Health care workers typically don gloves <strong>and</strong><br />
gowns when entering the room <strong>of</strong> an infected or colonized patient, <strong>and</strong><br />
remove them upon exiting (followed immediately by h<strong>and</strong>washing) to reduce<br />
the likelihood <strong>of</strong> clothing or equipment contamination that could transmit<br />
pathogens to other patients;<br />
• Use <strong>of</strong> dedicated or disposable examining equipment for patients colonized or<br />
infected with VRE <strong>and</strong>/or C. difficile: Hospital equipment (ie, blood pressure<br />
cuffs, thermometers) remains in a patient’s room <strong>and</strong> is not carried from room<br />
to room; <strong>and</strong><br />
• <strong>Patient</strong> <strong>and</strong>/or staff cohorting for patients colonized or infected with VRE<br />
<strong>and</strong>/or C. difficile: <strong>Patient</strong>s colonized or infected with similar pathogens are<br />
admitted to specific floors <strong>of</strong> the hospital where designated health care<br />
workers care only for patients colonized or infected with these pathogens.<br />
127
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Nosocomial infections, including C. difficile-associated diarrhea <strong>and</strong> VRE, significantly<br />
increase the morbidity <strong>and</strong> mortality <strong>of</strong> hospitalized patients. 5,6 Both infections are also<br />
associated with increased hospital costs. Recent evidence also suggests there may be a<br />
relationship between C. difficile <strong>and</strong> VRE, with C. difficile infection identified as a risk factor for<br />
VRE infection. 7 The increased incidence <strong>of</strong> both VRE <strong>and</strong> C. difficile can be attributed to spread<br />
from patient to patient. 5,6 Failure to recognize these dissemination patterns may result in an<br />
inability to contain outbreaks when they occur in the hospital.<br />
C. difficile has been identified as the major, if not only, important cause <strong>of</strong> infectious<br />
diarrhea that develops in patients after hospitalization, occurring in up to 30% <strong>of</strong> adult<br />
hospitalized patients who developed diarrhea. 5 One study found an acquisition rate <strong>of</strong> 13% for<br />
patients hospitalized 1-2 weeks, which increased to 50% for patients hospitalized >4 weeks. 8 In<br />
addition, the incidence <strong>of</strong> C. difficile infection has increased in recent years, with one study<br />
reporting a 5-fold increase in clinical infection between 1993 <strong>and</strong> 1996. 9 C. difficile infection<br />
increases lengths <strong>of</strong> stay, <strong>of</strong>ten to as long as 18-30 days 10,11 <strong>and</strong>, when fulminant, can lead to<br />
exploratory <strong>and</strong> therapeutic surgical procedures. 12 Mortality attributable to C. difficile, while<br />
reported, occurs in fewer than 5% <strong>of</strong> patients. 13 The costs associated with C. difficile diarrhea,<br />
while not well described, may be as high as $10,000 per patient. 14<br />
VRE, first described in 1988, currently accounts for greater than 25% <strong>of</strong> all nosocomial<br />
enterococci. 6 Early national data suggested that infections with VRE were associated with<br />
mortality rates <strong>of</strong> over 36%, more than double that <strong>of</strong> patients with vancomycin-susceptible<br />
(VSE) infections. 15 While later studies called some <strong>of</strong> these results into question, 16,17 the most<br />
recent studies have again suggested that vancomycin-resistance carries an independent effect on<br />
mortality. 18 VRE infections are also associated with significantly higher hospital costs than those<br />
due to VSE. 18<br />
Although C. difficile <strong>and</strong> VRE are among the most common nosocomial pathogens that<br />
have significant effects on morbidity, mortality, <strong>and</strong> cost, there are a number <strong>of</strong> other nosocomial<br />
pathogens which could also be studied. These include pathogens such as methicillin-resistant<br />
Staphylococcus aureus (MRSA), extended-spectrum beta-lactamase (ESBL) producing<br />
Enterobacteriaceae, Acinetobacter species, <strong>and</strong> Pseudomonas aeruginosa. While these are all<br />
important nosocomial pathogens, C. difficile <strong>and</strong> VRE were chosen as examples because they are<br />
extremely common, <strong>and</strong> they represent both antibiotic-susceptible (C. difficile) <strong>and</strong> antibioticresistant<br />
(VRE) pathogens. Additionally, (unlike MRSA <strong>and</strong> P. aeruginosa) the epidemiology <strong>of</strong><br />
both pathogens is complex, representing both person-to-person spread <strong>and</strong> association with prior<br />
antibiotic use, allowing for a more comprehensive discussion <strong>of</strong> the relative merits <strong>of</strong> both<br />
antimicrobial use interventions <strong>and</strong> barrier precaution interventions (see Chapter 15 for more<br />
discussion regarding other antimicrobial intervention practices) <strong>and</strong> their general application to<br />
other pathogens.<br />
Opportunities for Impact<br />
As noted above, both VRE <strong>and</strong> C. difficile affect a large proportion <strong>of</strong> hospitalized<br />
patients. Improvements in barrier precaution interventions against these pathogens would have a<br />
tremendous impact. There are few data regarding the percentage <strong>of</strong> hospitals that employ any<br />
one <strong>of</strong> a number <strong>of</strong> barrier precautions (eg, gowns, gloves, disposable thermometers). 19 In<br />
addition, while st<strong>and</strong>ard practice is to apply barrier precautions for patients with nosocomial<br />
pathogens with demonstrated horizontal spread, compliance with precautions is frequently<br />
128
poor, 20 <strong>of</strong>ten below 50%. 21 Purported reasons for this lack <strong>of</strong> compliance include lack <strong>of</strong><br />
resources <strong>and</strong> busy staff workload. 20 Regardless, these results suggest that the opportunity for<br />
improvement in these practices is great.<br />
Study Designs<br />
A structured search <strong>of</strong> the PubMed database (including MEDLINE) <strong>and</strong> review <strong>of</strong> the<br />
bibliographies <strong>of</strong> relevant articles identified 19 studies that have examined the implementation <strong>of</strong><br />
barrier precaution practices designed to impact the incidence <strong>of</strong> VRE <strong>and</strong>/or C. difficile infection<br />
(Table 13.1, 13.2, 13.3). All studies found on literature search were included in this review<br />
except for those reporting very small outbreaks (defined as fewer than 10 cases <strong>of</strong> C. difficile or<br />
VRE). Sixteen <strong>of</strong> the reviewed studies were before-after observational cohort studies (Level 3),<br />
in which baseline data regarding incidence <strong>of</strong> VRE or C. difficile were obtained during an<br />
observational period <strong>and</strong> compared to a second period after implementation <strong>of</strong> an intervention.<br />
Crude comparability data on the before <strong>and</strong> after groups (eg, total admissions, patient census)<br />
were provided in 2 reports 22,23 while only one study statistically compared the before <strong>and</strong> after<br />
groups to assess comparability. 24 Three reports 25-27 detailed unblinded comparative studies<br />
(Level 2) in which patients on different wards were assigned different interventions. Each <strong>of</strong><br />
these studies assessed the comparability <strong>of</strong> the study groups on the basis <strong>of</strong> underlying<br />
demographic variables.<br />
Study Outcomes<br />
All <strong>of</strong> the studies reviewed reported changes in the incidence or prevalence <strong>of</strong> either VRE<br />
or C. difficile as a result <strong>of</strong> barrier precaution interventions (Level 1). For studies investigating C.<br />
difficile, all outcomes were reported in terms <strong>of</strong> clinical infections. For studies investigating<br />
VRE, outcomes were reported as VRE colonization <strong>and</strong>/or infection rates.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
As both VRE <strong>and</strong> C. difficile have clearly been shown to be transferable from patient-topatient,<br />
interventions designed to improve barrier precautions yield significant reductions in the<br />
incidence <strong>of</strong> infection with these two pathogens. All studies that examined the effect <strong>of</strong> enhanced<br />
barrier precautions on C. difficile infection demonstrated benefit, suggesting that barrier<br />
precaution interventions are effective in controlling its emergence. Most studies employed a<br />
multifaceted approach including several different barrier precaution components. For example,<br />
one study combined use <strong>of</strong> vinyl gloves <strong>and</strong> ongoing educational interventions, 26 another<br />
included cohorting, culture screening, <strong>and</strong> daily room disinfection, 28 while another combined<br />
reinforcement <strong>of</strong> enteric precautions, replacement <strong>of</strong> electronic thermometers, <strong>and</strong> institution <strong>of</strong><br />
closed paper towel dispensers. 29 Given the varied components <strong>of</strong> barrier precaution interventions<br />
instituted in different studies, it is difficult to determine the specific effect <strong>of</strong> any individual<br />
component.<br />
The evidence <strong>of</strong> effectiveness <strong>of</strong> barrier precautions for VRE is somewhat less clear-cut.<br />
All but 4 27,30-32 <strong>of</strong> the studies examining the effect <strong>of</strong> barrier precautions on VRE demonstrated a<br />
benefit, but study design differences <strong>and</strong> particular epidemiologic trends may account for the<br />
inconsistent findings.<br />
One <strong>of</strong> the 4 studies that noted no significant effect compared glove use to glove <strong>and</strong><br />
gown use. 27 The second 30 noted that the emergence <strong>of</strong> VRE at the study institution was due to<br />
multiple genetically-unrelated strains, suggesting that person-to-person spread was less<br />
important at that site. It is thus not surprising that barrier precautions would have less <strong>of</strong> an<br />
129
effect. In the third study, 32 routine rectal swab surveillance <strong>and</strong> contact precautions were<br />
instituted in response to a clinical outbreak <strong>of</strong> VRE <strong>and</strong> surveillance was continued for only 6<br />
months. Since surveillance was not conducted prior to institution <strong>of</strong> precautions, it is impossible<br />
to say what the colonization prevalence had been prior to the intervention. Furthermore, as the<br />
authors point out, it may be that the outbreak would have been much worse had the precautions<br />
not been put in place. Finally, no determination <strong>of</strong> genetic relatedness (<strong>and</strong> hence spread) was<br />
made in this study. In the fourth study, 31 while there was a reduction in the isolation <strong>of</strong> VRE,<br />
there was not complete eradication. According to the authors, the most likely reason for this lessthan-optimal<br />
response was poor compliance with contact precaution guidelines.<br />
Thus, it appears that enhanced barrier precautions are generally effective in reducing the<br />
incidence <strong>of</strong> VRE but that various aspects <strong>of</strong> both the epidemiology <strong>of</strong> the VRE outbreak <strong>and</strong> the<br />
implementation <strong>of</strong> guidelines may temper the effectiveness <strong>of</strong> interventions. Similar to the<br />
studies investigating response <strong>of</strong> C. difficile to barrier precautions, most studies <strong>of</strong> VRE<br />
employed several components <strong>of</strong> barrier precautions as part <strong>of</strong> a multifaceted approach (Table<br />
13.1). It is thus difficult to determine the specific effect <strong>of</strong> any individual component.<br />
Potential for Harm<br />
None <strong>of</strong> the reviewed studies reported any assessment <strong>of</strong> possible harm as a result <strong>of</strong> the<br />
barrier precaution interventions. In fact, the implementation <strong>of</strong> barrier precautions is unlikely to<br />
result in harm to the patient. One potential concern is that time necessary to comply with the<br />
interventions (eg, gowning, gloving), might make health care workers less likely to complete<br />
tasks necessary to provide acceptable patient care. Indeed, it has recently been noted that health<br />
care workers were half as likely to enter the rooms <strong>of</strong> patients on contact isolation. 33<br />
Furthermore, while contact precautions appeared to have little effect on patient examination by<br />
resident physicians, attending physicians were 50% less likely to examine a patient on contact<br />
precautions compared to a patient not on precautions. 34 Future studies should address these<br />
concerns by documenting the time required to adhere to barrier precautions, <strong>and</strong> determining the<br />
potential impact <strong>of</strong> precautions on patient care.<br />
Another potentially harmful consequence <strong>of</strong> barrier precaution interventions is the<br />
psychological effect that contact precautions may have on the isolated patient. While research<br />
has examined the effects <strong>of</strong> sensory deprivation <strong>and</strong> social isolation, a recent review <strong>of</strong> the<br />
literature noted little progress in the investigation <strong>of</strong> the psychological effects <strong>of</strong> contact<br />
isolation. 35<br />
Costs <strong>and</strong> Implementation<br />
It seems apparent that the more complicated an intervention, the less likely health care<br />
workers will adhere to it. While 2 studies noted compliance with barrier precautions at close to<br />
90%, 21,24 others noted levels closer to 70%. 31 One study actually noted compliance levels to be<br />
significantly higher in those health care workers who used both gowns <strong>and</strong> gloves compared to<br />
those using only gowns. 27 This somewhat counterintuitive finding suggests that other factors<br />
may be at play in influencing compliance. Of the reviewed studies that reported compliance<br />
levels, all did so relatively shortly after the initial implementation <strong>of</strong> interventions. Future studies<br />
should assess compliance with guidelines over a longer period.<br />
Four studies reported the costs <strong>of</strong> specific interventions. Implementation <strong>of</strong> use <strong>of</strong><br />
disposable thermometers was estimated at $14,055 per year at a 343-bed institution. 22 Another<br />
study <strong>of</strong> the impact <strong>of</strong> disposable thermometers estimated that the cost per prevented C. difficile<br />
infection would be approximately $611. 25 A study using a multifaceted approach estimated that<br />
130
the annual expenses due directly to increased dem<strong>and</strong> for gowns <strong>and</strong> gloves were approximately<br />
$11,000. 31 Finally, a multifaceted intervention at a 254-bed long-term care facility which<br />
included education, gowns <strong>and</strong> gloves for resident contact, no sharing <strong>of</strong> personal equipment,<br />
<strong>and</strong> daily double cleaning <strong>of</strong> resident rooms <strong>and</strong> wheelchairs, estimated the total cost <strong>of</strong> the<br />
intervention to be $12,061 Canadian (approximately $8000 US). 36<br />
The costs <strong>of</strong> implementing a program to enhance barrier precaution practices must be<br />
balanced against the potential cost savings due to decreased incidence <strong>of</strong> nosocomial infections.<br />
Both VRE <strong>and</strong> C. difficile infections have been associated with significantly increased length <strong>of</strong><br />
hospital stay. 5,6 Preventing even a small number <strong>of</strong> these infections is likely to have a significant<br />
financial impact. While several <strong>of</strong> the reviewed studies documented costs associated with<br />
various interventions, 22,25,26,31,36 no study systematically compared these costs to the potential<br />
cost savings <strong>of</strong> infections prevented.<br />
Comment<br />
The majority <strong>of</strong> reviewed studies demonstrated a significant reduction in the incidence <strong>of</strong><br />
VRE or C. difficile following barrier precaution interventions. The fact that not all studies found<br />
a benefit suggests that future studies should identify those scenarios (eg, outbreak, endemic<br />
colonization, etc.) in which attention to barrier precautions is most likely to be beneficial. In<br />
addition, it is possible that a combined intervention involving both enhanced barrier precautions<br />
as well as antibiotic formulary interventions might be needed in order to effect the greatest<br />
possible change in VRE <strong>and</strong> C. difficile infection rates. While these studies, much like those that<br />
examined the impact <strong>of</strong> antibiotic use practices, demonstrated short-term success, future studies<br />
should determine the efficacy <strong>of</strong> such interventions over the long term. Finally, the costeffectiveness<br />
<strong>of</strong> such strategies should be investigated.<br />
131
Table 13.1. Studies <strong>of</strong> multifaceted approaches with <strong>and</strong> without “cohorting”*<br />
Study Setting Compliance Study<br />
Design,<br />
Outcomes<br />
725-bed academic medical center in<br />
Philadelphia in 1987-88: beforeafter<br />
study <strong>of</strong> impact <strong>of</strong><br />
multifaceted intervention (isolation<br />
precautions, clindamycin<br />
restriction) on C. difficile 37<br />
350-bed acute care hospital in<br />
Virginia in 1987-96: before-after<br />
study <strong>of</strong> impact <strong>of</strong> multifaceted<br />
intervention on C. difficile<br />
infections 23<br />
840-bed tertiary care center in<br />
Brussels in 1989-90: impact <strong>of</strong> a<br />
multifaceted infection control<br />
intervention, including cohorting,<br />
on incidence <strong>of</strong> C. difficile 28<br />
Bone marrow transplant unit <strong>of</strong> a<br />
large academic medical center in<br />
Texas in 1995: impact <strong>of</strong><br />
multifaceted infection control<br />
intervention on C. difficile attack<br />
rate 29<br />
Tertiary-care Veterans Affairs<br />
<strong>Medical</strong> Center in Brooklyn in<br />
1991-95: impact <strong>of</strong> multifaceted<br />
infection control intervention on<br />
VRE rates 30<br />
22-bed oncology unit in a 650-bed<br />
tertiary care hospital in New York<br />
in 1993-95: impact <strong>of</strong> multifaceted<br />
infection control program, including<br />
cohorting, on VRE infection <strong>and</strong><br />
colonization 24<br />
375-bed community hospital in<br />
Indianapolis in 1995-96: impact <strong>of</strong><br />
cohorting on VRE prevalence 21<br />
NA Level 3,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
91.7% <strong>of</strong><br />
persons who<br />
entered room<br />
used gowns<br />
<strong>and</strong> gloves<br />
appropriately<br />
Compliance<br />
with recommendations<br />
rose from<br />
22% to 88%<br />
(p
254-bed long-term care facility in<br />
Toronto in 1996-97: impact <strong>of</strong><br />
barrier precautions including<br />
cohorting on prevalence <strong>of</strong> VRE 36<br />
NA Level 3,<br />
Level 1<br />
133<br />
4/85 (4.7%) patients initially<br />
screened were VRE colonized.<br />
No patients in subsequent<br />
screenings were positive.
Table 13.1. Studies <strong>of</strong> multifaceted approaches with <strong>and</strong> without “cohorting” (cont.)<br />
23-bed oncology unit in a 1300bed<br />
teaching hospital in South<br />
Africa in 1998: impact <strong>of</strong> barrier<br />
precautions including cohorting<br />
on VRE prevalence 39<br />
347-bed tertiary care medical<br />
center in Massachusetts in 1993:<br />
impact <strong>of</strong> a multifaceted infection<br />
control intervention including<br />
cohorting on VRE infection <strong>and</strong><br />
colonization 31<br />
NA Level 3,<br />
Level 1<br />
Overall<br />
h<strong>and</strong>washing<br />
complianc<br />
e was 71%<br />
134<br />
Level 3,<br />
Level 1<br />
* NA indicates not applicable; VRE, vancomycin-resistant enterococci.<br />
VRE colonization decreased from<br />
19/34 (55%) patients to 1/14 (7%)<br />
following implementation <strong>of</strong><br />
infection control interventions<br />
In the year prior interventions,<br />
116 patients were colonized or<br />
infected with VRE, compared<br />
with 126 in the year after<br />
implementation.
Table 13.2. Studies <strong>of</strong> barrier precaution interventions*<br />
Study Setting Compliance Study<br />
Design,<br />
Outcomes<br />
370-bed academic medical center in<br />
Massachusetts in 1991-92: beforeafter<br />
study <strong>of</strong> impact <strong>of</strong> infection<br />
control interventions on C. difficile<br />
incidence 38<br />
Veterans Administration <strong>Medical</strong><br />
Center in Minnesota in 1986-87:<br />
impact <strong>of</strong> universal glove use on<br />
incidence <strong>of</strong> C. difficile 26<br />
8-bed combined medical <strong>and</strong> surgical<br />
ICU in a 235-bed acute care hospital<br />
in New York City in 1990-91: impact<br />
<strong>of</strong> barrier precautions on VRE<br />
colonization 1<br />
250-bed university-affiliated hospital<br />
in Rhode Isl<strong>and</strong> in 1991-92: impact<br />
<strong>of</strong> sequential barrier precaution<br />
intervention on VRE 40<br />
181 consecutive patients admitted to<br />
the medical ICU in a 900-bed urban<br />
teaching hospital in Chicago in 1994-<br />
95: comparison <strong>of</strong> impact <strong>of</strong> gown<br />
<strong>and</strong> glove vs. glove on incidence <strong>of</strong><br />
VRE colonization 27<br />
550-bed tertiary teaching hospital in<br />
Minneapolis in 1993-94: impact <strong>of</strong><br />
barrier precautions on VRE<br />
colonization 32<br />
NA Level 3,<br />
Level 1<br />
Mean glove<br />
use/100 pts:<br />
4539 on glove<br />
ward; 3603 on<br />
control ward<br />
(p=NS)<br />
135<br />
Level 2,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
NA Level 3,<br />
Level 1<br />
Compliance in<br />
glove <strong>and</strong><br />
gown group,<br />
79%; glove<br />
group, 62%<br />
(p
Table 13.3. Studies <strong>of</strong> use <strong>of</strong> dedicated or disposable examining equipment*<br />
Study Setting Complianc<br />
e<br />
343-bed acute hospital <strong>and</strong><br />
538-bed skilled nursing<br />
facility in New York: beforeafter<br />
study <strong>of</strong> impact <strong>of</strong><br />
replacing electronic<br />
thermometers with disposable<br />
thermometers on C. difficile<br />
infection rate 22<br />
20 inpatient units in a 700-bed<br />
university hospital in<br />
Virginia: r<strong>and</strong>omized<br />
crossover trial <strong>of</strong> impact <strong>of</strong><br />
disposable thermometers for<br />
prevention <strong>of</strong> C. difficile 25<br />
343-bed acute care facility in<br />
New York in 1992: impact <strong>of</strong><br />
change to tympanic<br />
thermometers on VRE<br />
incidence 22<br />
100%<br />
replacemen<br />
t <strong>of</strong><br />
electronic<br />
thermometers<br />
100%<br />
compliance<br />
with use <strong>of</strong><br />
specific<br />
types <strong>of</strong><br />
thermometers<br />
100%<br />
switch to<br />
tympanic<br />
thermometers<br />
Study<br />
Design,<br />
Outcomes<br />
Level 3,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
136<br />
Change in C. difficile or VRE<br />
Incidence <strong>of</strong> C. difficile decreased<br />
from 2.71 to 1.76 cases per 1000<br />
patients in the acute hospital<br />
(p
References<br />
1. H<strong>and</strong>werger S, Raucher B, Altarac D, Monka J, Marchione S, Singh KV, et al. Nosocomial<br />
outbreak due to Enterococcus faecium highly resistant to vancomycin, penicillin, <strong>and</strong><br />
gentamicin. Clin Infect Dis. 1993;16:750-755.<br />
2. Chang VT, Nelson K. The role <strong>of</strong> physical proximity in nosocomial diarrhea. Clin Infect<br />
Dis. 2000;31:717-722.<br />
3. Byers KE, Durbin LJ, Simonton BM, Anglim AM, Adal KA, Farr BM. Disinfection <strong>of</strong><br />
hospital rooms contaminated with vancomycin-resistant Enterococcus faecium. Infect<br />
Control Hosp Epidemiol. 1998;19:261-264.<br />
4. Mayfield JL, Leet T, Miller J, Mundy LM. Environmental control to reduce transmission <strong>of</strong><br />
Clostridium difficile. Clin Infect Dis. 2000;31:995-1000.<br />
5. Johnson S, Gerding DN. Clostridium difficile-associated diarrhea. Clin Infect Dis.<br />
1998;26:1027-1036.<br />
6. Murray BE. Vancomycin-resistant enterococcal infections. N Engl J Med. 2000;342:710-<br />
721.<br />
7. Roghmann MC, McCarter RJ, Brewrink J, Cross AS, Morris JG. Clostridium difficile<br />
infection is a risk factor for bacteremia due to vancomycin-resistant enterococci (VRE) in<br />
VRE-colonized patients with acute leukemia. Clin Infect Dis. 1997;25:1056-1059.<br />
8. Clabots CR, Johnson S, Olson MM, Peterson LR, Gerding DN. Acquisition <strong>of</strong> Clostridium<br />
difficile by hospitalized patients: evidence <strong>of</strong> colonized new admissions as the source <strong>of</strong><br />
infection. J Infect Dis. 1992;166:561-567.<br />
9. Wilcox MH, Smyth ETM. Incidence <strong>and</strong> impact <strong>of</strong> Clostridium difficile infection in the<br />
UK, 1993-1996. J Hosp Infect. 1998;39:181-187.<br />
10. MacGowan AP, Brown I, Feeney R, Lovering A, McCulloch SY, Reeves DS. Clostridium<br />
difficile associated diarrhea <strong>and</strong> length <strong>of</strong> hospital stay. J Hosp Infect. 1995;31:241-244.<br />
11. Riley TV, Codde JP, Rouse IL. Increase length <strong>of</strong> stay due to Clostridium difficile<br />
associated diarrhoea. Lancet. 1995;345:455-456.<br />
12. Kent KC, Rubin MS, Wroblewski L, Hanff PA, Sline W. The impact <strong>of</strong> Clostridium<br />
difficile on a surgical service. Ann Surg. 1998;227:296-301.<br />
13. Olson MM, Shanholtzer CJ, Lee JT, Gerding DN. Ten years <strong>of</strong> prospective Clostridium<br />
difficile-associated disease surveillance <strong>and</strong> treatment at the Minneapolis VA <strong>Medical</strong><br />
Center, 1982-1991. Infect Control Hosp Epidemiol. 1994;15:371-381.<br />
14. Spencer RC. Clinical impact <strong>and</strong> associated costs <strong>of</strong> Clostridium difficile-associated<br />
disease. J Antimicrob Chemother. 1998;41(Suppl C):C5-12.<br />
15. Centers for Disease Control <strong>and</strong> Prevention (CDC). Nosocomial enterococci resistant to<br />
vancomycin - United States, 1989-93. MMWR Morb Mortal Wkly Rep. 1993;42:579-597.<br />
16. Lautenbach E, Bilker WB, Brennan PJ. Enterococcal bacteremia: risk factors for<br />
vancomycin resistance <strong>and</strong> predictors <strong>of</strong> mortality. Infect Control Hosp Epidemiol.<br />
1999;20:318-323.<br />
17. Linden PK, Pasculle AW, Manez R, Kramer DJ, Fung JJ, Pinna AD, et al. Differences in<br />
outcomes for patients with bacteremia due to vancomycin-resistant Enterococcus faecium<br />
or vancomycin-susceptible E. faecium. Clin Infect Dis. 1996;22:663-670.<br />
18. Stosor V, Peterson LR, Postelnick M, Noskin GA. Enterococcus faecium bacteremia: does<br />
vancomycin resistance make a difference? Arch Intern Med. 1998;158:522-527.<br />
19. Scheckler WE, Brimhall D, Buck AS, Farr BM, Friedman C, Garibaldi RA, et al.<br />
Requirements for infrastructure <strong>and</strong> essential activities <strong>of</strong> infection control <strong>and</strong><br />
137
epidemiology in hospitals: a consensus panel report. Infect Control Hosp Epidemiol.<br />
1998;19:114-124.<br />
20. Kollef MH, Fraser VJ. Antibiotic resistance in the intensive care unit. Ann Intern Med.<br />
2001;134:298-314.<br />
21. Jochimsen EM, Fish L, Manning K, Young S, Singer DA, Baker R, et al. Control <strong>of</strong><br />
vancomycin-resistant enterococci at a community hospital: efficacy <strong>of</strong> patient <strong>and</strong> staff<br />
cohorting. Infect Control Hosp Epidemiol. 1999;20:106-109.<br />
22. Brooks SE, Veal RO, Kramer M, Dore L, Schupf N, Adachi M. Reduction in the incidence<br />
<strong>of</strong> Clostridium difficile-associated diarrhea in an acute care hospital <strong>and</strong> a skilled nursing<br />
facility following replacement <strong>of</strong> electronic thermometers with single-use disposables.<br />
Infect Control Hosp Epidemiol. 1992;13:98-103.<br />
23. Zafar AB, Gaydos LA, Furlong WB, Nguyen MH, Mennonna PA. Effectiveness <strong>of</strong><br />
infection control program in controlling nosocomial Clostridium difficile. Am J Infect<br />
Control. 1998;26:588-593.<br />
24. Montecalvo MA, Jarvis WR, Uman J, Shar DK, Petrullo C, Rodney K, et al. Infectioncontrol<br />
measures reduce transmission <strong>of</strong> vancomycin-resistant enterococci in an endemic<br />
setting. Ann Intern Med. 1999;131:269-272.<br />
25. Jernigan JA, Siegman-Igra Y, Guerrant RC, Farr BM. A r<strong>and</strong>omized crossover study <strong>of</strong><br />
disposable thermometers for prevention <strong>of</strong> Clostridium difficile <strong>and</strong> other nosocomial<br />
infections. Infect Control Hosp Epidemiol. 1998;19:494-499.<br />
26. Johnson S, Gerding DN, Olson MM, Weiler MD, Hughes RA, Clabots CR, et al.<br />
Prospective, controlled study <strong>of</strong> vinyl glove use to interrupt Clostridium difficile<br />
nosocomial transmission. Am J Med. 1990;88:137-140.<br />
27. Slaughter S, Hayden MK, Nathan C, Hu TC, Rice T, Van Voorhis J, et al. A comparison <strong>of</strong><br />
the effect <strong>of</strong> universal use <strong>of</strong> gloves <strong>and</strong> gowns with that <strong>of</strong> glove use alone on acquisition<br />
<strong>of</strong> vancomycin-resistant enterococci in a medical intensive care unit. Ann Intern Med.<br />
1996;125:448-456.<br />
28. Struelens MJ, Maas A, Nonh<strong>of</strong>f C, Deplano A, Rost F, Serruys E, et al. Control <strong>of</strong><br />
nosocomial transmission <strong>of</strong> Clostridium difficile based on sporadic case surveillance. Am J<br />
Med. 1991;91(Suppl 3B):138S-44S.<br />
29. Hanna H, Raad I, Gonzalez V, Umphrey J, Tarr<strong>and</strong> J, Neumann J, et al. Control <strong>of</strong><br />
nosocomial Clostridium difficile transmission in bone marrow transplant patients. Infect<br />
Control Hosp Epidemiol. 2000;21:226-8.<br />
30. Quale J, L<strong>and</strong>man D, Atwood E, Kreiswirth B, Willey BM, Ditore V, et al. Experience<br />
with a hospital-wide outbreak <strong>of</strong> vancomycin-resistant enterococci. Am J Infect Control.<br />
1996;24:372-379.<br />
31. Lai KK, Kelley AL, Melvin ZS, Belliveau PP, Fontecchio SA. Failure to eradicate<br />
vancomycin-resistant enterococci in a university hospital <strong>and</strong> the cost <strong>of</strong> barrier<br />
precautions. Infect Control Hosp Epidemiol. 1998;19:647-652.<br />
32. Wells CL, Juni BA, Cameron SB, Mason KR, Dunn DL, Ferrieri P, et al. Stool carriage,<br />
clinical isolation, <strong>and</strong> mortality during an outbreak <strong>of</strong> vancomycin-resistant enterococci in<br />
hospitalized medical <strong>and</strong>/or surgical patients. Clin Infect Dis. 1995;21:45-50.<br />
33. Kirkl<strong>and</strong> KB, Weinstein JM. Adverse effects <strong>of</strong> contact isolation. Lancet. 1999;354:1177-<br />
1178.<br />
34. Higgins LA, Saint S, Nallamothu BK, Chenoweth C. Do physicians examine patients under<br />
contact precautions less frequently? Paper presented at: 24th Annual Meeting <strong>of</strong> the<br />
Society for General Internal Medicine; May 2-5, 2001; San Diego, CA.<br />
138
35. Gammon J. The psychological consequences <strong>of</strong> source isolation: a review <strong>of</strong> the literature.<br />
J Clin Nurs. 1999;8:13-21.<br />
36. Armstrong-Evans M, Litt M, McArthur MA, Willey B, Cann D, Liska S, et al. Control <strong>of</strong><br />
transmission <strong>of</strong> vancomycin-resistant Enterococcus faecium in a long-term care facility.<br />
Infect Control Hosp Epidemiol. 1999;20:312-317.<br />
37. Brown E, Talbot GH, Axelrod P, Provencher M, Hoegg C. Risk factors for Clostridium<br />
difficile toxin-associated diarrhea. Infect Control Hosp Epidemiol. 1990;11:283-290.<br />
38. Lai KK, Melvin ZS, Menard MJ, Kotilainen HR, Baker S. Clostridium difficile-associated<br />
diarrhea: epidemiology, risk factors, <strong>and</strong> infection control. Infect Control Hosp Epidemiol.<br />
1997;18:628-632.<br />
39. McCarthy KM, Van Nierop W, Duse A, Von Gottberg A, Kassel M, Perovic O, et al.<br />
Control <strong>of</strong> an outbreak <strong>of</strong> vancomycin-resistant Enterococcus faecium in an oncology ward<br />
in South Africa: effective use <strong>of</strong> limited resources. J Hosp Infect. 2000;44:294-300.<br />
40. Boyce JM, Opal SM, Chow JW, Zervos MJ, Potter-Bynoe G, Sherman CB, et al. Outbreak<br />
<strong>of</strong> multidrug-resistant Enterococcus faecium with transferable vanB class vancomycin<br />
resistance. J Clin Microbiol. 1994;32:1148-53.<br />
139
140
Chapter 14. Impact <strong>of</strong> Changes in Antibiotic Use <strong>Practices</strong> on Nosocomial<br />
Infections <strong>and</strong> Antimicrobial Resistance – Clostridium Difficile <strong>and</strong><br />
Vancomycin-resistant Enterococcus (VRE)<br />
Ebbing Lautenbach, MD, MPH, MSCE<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Background<br />
As discussed in the chapters on h<strong>and</strong>washing <strong>and</strong> barrier precautions (Chapters 12 <strong>and</strong><br />
13), hospital infection control has historically focused on preventing the transmission <strong>of</strong><br />
nosocomial pathogens—either from patient to patient or from provider to patient. The potential<br />
role <strong>of</strong> overseeing hospital-wide antibiotic use as an infection control measure has also been<br />
recognized for many years. 1 With the widespread emergence <strong>of</strong> nosocomial antibiotic-resistant<br />
infections over the past 10-15 years, institutional efforts to control antibiotic use have become a<br />
priority for infection control. 2,3<br />
The practices reviewed in this chapter involve institutional efforts to control antibiotic<br />
use as a means <strong>of</strong> controlling complications <strong>of</strong> antibiotic overuse or misuse. In evaluating the<br />
potential benefits <strong>of</strong> these practices, the focus is on the impacts <strong>of</strong> antibiotic use on infections<br />
with vancomycin-resistant enterococci (VRE) 4 <strong>and</strong> Clostridium difficile. 5 These pathogens<br />
represent two <strong>of</strong> the most important nosocomial pathogens with relationships to inappropriate<br />
antibiotic use. Moreover, as suggested by recent evidence, infection with C. difficile may<br />
represent a risk factor for infection with VRE. 6<br />
Practice description<br />
Interventions designed to limit the use <strong>of</strong> antibiotics may take many forms. Specific<br />
practices reviewed in the chapter include:<br />
• Infectious diseases physician approval 7 – all requests for an antibiotic are<br />
discussed with an infectious diseases physician who decides whether use is<br />
appropriate<br />
• Monitoring <strong>of</strong> antibiotic use by pharmacy service 8 – pharmacists monitor the<br />
use <strong>of</strong> certain antibiotics <strong>and</strong> make recommendations for changes to the<br />
prescriber<br />
• Guidelines for antimicrobial use 8 – dissemination to physicians <strong>of</strong> guidelines<br />
describing appropriate <strong>and</strong> inappropriate use<br />
• Therapeutic substitution 9 – use <strong>of</strong> one agent replaced by another agent with<br />
similar spectrum <strong>of</strong> activity<br />
• Computer-assisted prescribing 10 – computer-based restriction <strong>of</strong> agent with<br />
extra prompts requesting documentation <strong>of</strong> indication for agent<br />
• Antibiotic-management program (AMP) 11 – continuation <strong>of</strong> antibiotic after a<br />
specific duration requires approval from either an infectious diseases<br />
physician or pharmacist on the AMP<br />
141
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
This chapter focuses on 2 <strong>of</strong> the most important nosocomial pathogens: VRE <strong>and</strong> C.<br />
difficile. VRE currently accounts for greater than 25% <strong>of</strong> all nosocomial enterococci 4 <strong>and</strong> confers<br />
an increased risk <strong>of</strong> death, independent <strong>of</strong> comorbid conditions that may have initially led to the<br />
infection. 12 VRE infections are also associated with significantly higher hospital costs than those<br />
due to vancomycin-sensitive enterococci (VSE) 12 (see Chapter 13). C. difficile represents the<br />
major, if not only, important infectious cause <strong>of</strong> nosocomial diarrhea. 5 Although death<br />
attributable to C. difficile occurs in less than 5% <strong>of</strong> patients, 17 the impact <strong>of</strong> C. difficile infection<br />
remains significant. <strong>Patient</strong>s may require substantially longer lengths <strong>of</strong> hospital stay—upwards<br />
<strong>of</strong> 18-30 days, 18,19 with exploratory <strong>and</strong> therapeutic surgical procedures required in severe<br />
cases. 20 It has also been suggested that more debilitated patients (eg, in rehabilitation centers or<br />
long-term care facilities) may be at even greater risk for increased morbidity <strong>and</strong> mortality due to<br />
C. difficile infection. 21 The costs associated with C. difficile diarrhea, while not well described,<br />
are estimated to be as high as $10,000 per patient 22 (see Chapter 13).<br />
Opportunities for Impact<br />
Over half <strong>of</strong> all hospitalized patients are treated with antibiotics. 23 The antibiotics<br />
represent a significant portion <strong>of</strong> overall health care costs, accounting for between 20% <strong>and</strong> 50%<br />
<strong>of</strong> total hospital drug expenditures. 23 It has been estimated that 50% <strong>of</strong> all antibiotics prescribed<br />
are either at the wrong dose, the wrong drug, or taken for the wrong duration. 24,25 These findings<br />
suggest that there is significant room for improvement in antibiotic prescribing practices.<br />
Most hospitals employ formulary restrictions for certain medications (particularly<br />
expensive agents, selecting one drug from a group <strong>of</strong> equivalent agents). However, only a<br />
minority <strong>of</strong> hospitals uses formulary restrictions to limit the use <strong>of</strong> entire antibiotic classes or<br />
specific agents. Those hospitals that do employ antimicrobial formulary restrictions most <strong>of</strong>ten<br />
do so as a means <strong>of</strong> controlling costs, rather than as an infection control measure. 26 Thus, there<br />
remain substantial opportunities to exp<strong>and</strong> upon these existing formulary programs to control the<br />
emergence <strong>of</strong> antimicrobial resistance.<br />
Study Designs<br />
A structured search <strong>of</strong> the PubMed database (including MEDLINE) <strong>and</strong> review <strong>of</strong> the<br />
bibliographies <strong>of</strong> relevant articles identified 10 studies that have examined methods to change<br />
antibiotic use with respect to VRE <strong>and</strong>/or C. difficile infection (Table 14.1). All <strong>of</strong> these studies<br />
were before-after observational cohort studies (Level 3) in which baseline data regarding<br />
incidence <strong>of</strong> VRE or C. difficile were obtained during an observational period <strong>and</strong> compared<br />
with a second time period after an intervention had been implemented. Data on baseline<br />
comparability <strong>of</strong> the before <strong>and</strong> after groups were not reported in 6 studies. 8-11,21,27 Two studies<br />
only reported similar admission <strong>and</strong> census rates during the before <strong>and</strong> after time periods, 7,28<br />
while 2 studies compared patients in the 2 time periods on the basis <strong>of</strong> numerous variables. 29,30<br />
Study Outcomes<br />
All <strong>of</strong> the studies reviewed reported changes in the clinical incidence or prevalence <strong>of</strong><br />
either VRE or C. difficile as a result <strong>of</strong> antibiotic practice interventions (Level 1). Studies<br />
investigating C. difficile measured clinical infections. Studies investigating VRE examined VRE<br />
infection 11 or VRE colonization. 8-10,27<br />
142
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Of the 10 studies listed in Table 14.1, all but 3 8,11,21 showed significant reductions in the<br />
incidence <strong>of</strong> C. difficile or VRE following practice changes. Several possibilities may explain the<br />
negative findings <strong>of</strong> these 3 studies. First, the interventions analyzed might not have produced<br />
significant alterations in antibiotic use, so that infection rates with the target pathogens remained<br />
unchanged. Second, it is possible that patient-to-patient spread <strong>of</strong> these pathogens limited the<br />
efficacy <strong>of</strong> the interventions, as this mode <strong>of</strong> transmission is well known to occur for both VRE<br />
<strong>and</strong> C. difficile, usually via the h<strong>and</strong>s <strong>of</strong> health care workers (see also Chapter 13). 31,32 Third,<br />
since environmental contamination occurs with both these pathogens, successful control <strong>of</strong> these<br />
organisms may require enhanced disinfection procedures in some cases. 33,34 Targeting antibiotic<br />
use may not be sufficient to reduce incidence <strong>of</strong> these pathogens since a significant number <strong>of</strong><br />
infected or colonized patients may serve as reservoirs. Under this scenario, the argument for<br />
barrier precautions as an adjunct measure to prevent spread <strong>of</strong> organisms from patient to patient<br />
becomes more persuasive (see Chapter 13). Indeed, although changes in antibiotic use practice<br />
were the primary intervention in all <strong>of</strong> the studies reviewed here, one study included a<br />
component <strong>of</strong> enhanced barrier precautions in the intervention. 21 Future studies should<br />
investigate the impact <strong>of</strong> such multifaceted interventions, both for VRE <strong>and</strong> C. difficile as well as<br />
for other nosocomial pathogens.<br />
Other Potential Benefits<br />
Although not the focus <strong>of</strong> this chapter, the practices reviewed here may have a beneficial<br />
impact on other emerging nosocomial pathogens strongly associated with inappropriate<br />
antibiotic use, such as extended-spectrum beta-lactamase (ESBL) producing<br />
Enterobacteriaceae. 35 In addition, although we have focused on control <strong>of</strong> VRE as an end in<br />
itself, a primary motivation to achieve this goal is the need to delay the emergence <strong>of</strong><br />
vancomycin-resistance in Staphylococcus aureus. 36,37 As S. aureus represents the most common<br />
nosocomial infection, 38 the development <strong>of</strong> high-level vancomycin resistance among<br />
staphylococci would constitute a public health disaster. 39<br />
Thus, practices that decrease the prevalence <strong>of</strong> VRE may play an important, albeit<br />
indirect, role in preventing or delaying this occurrence.<br />
Potential for Harm<br />
Few <strong>of</strong> the reviewed studies reported any assessment <strong>of</strong> possible harm as a result <strong>of</strong> the<br />
antibiotic use practice interventions. One potential result <strong>of</strong> interventions designed to reduce the<br />
use <strong>of</strong> one antibiotic, or antibiotic class, is the subsequent increase in the use <strong>of</strong> another class <strong>of</strong><br />
agents to compensate. In fact, one reviewed study 7 noted an increase in the use <strong>of</strong> other antianaerobic<br />
agents as clindamycin use decreased. Whether changes in antibiotic use results in<br />
changes in antimicrobial susceptibilities, either in the pathogen under study (eg, VRE, C.<br />
difficile) or in other nosocomial pathogens, it is a fertile ground for future study.<br />
Finally, efforts to decrease use <strong>of</strong> certain antibiotics might increase infection rates due to<br />
inappropriate withholding <strong>of</strong> appropriate antibiotics. However, one reviewed study 10 noted no<br />
increase in rates <strong>of</strong> surgical site infections following decrease in the use <strong>of</strong> vancomycin for<br />
preoperative prophylaxis (see also Subchapter 20.1).<br />
143
Costs <strong>and</strong> Implementation<br />
The costs <strong>of</strong> implementing a program to alter antibiotic use practices must be balanced<br />
against potential cost savings. Sources <strong>of</strong> savings may be reduced antibiotic use, use <strong>of</strong> less<br />
expensive agents rather than the more expensive newer agents, <strong>and</strong> potentially, reduced costs<br />
due to decreased incidence <strong>of</strong> nosocomial infections as a result <strong>of</strong> interventions. Although<br />
several studies reported cost savings due only to decreased antibiotic use, 10,11,29 analyses taking<br />
into account costs related to subsequent infections (or infections prevented) have been sparse.<br />
One study noted that cost savings from decreased use <strong>of</strong> clindamycin <strong>of</strong>fset the expenditures due<br />
to increased use <strong>of</strong> other antibiotics. 7 The authors suggested that if each case <strong>of</strong> C. difficile<br />
resulted in a cost <strong>of</strong> $2000, the savings to the hospital <strong>of</strong> the intervention could approach<br />
$162,000 annually based on the number <strong>of</strong> cases averted. 7<br />
Another cost <strong>of</strong> antibiotic use interventions is the expense <strong>of</strong> ongoing monitoring <strong>of</strong><br />
antibiotic use <strong>and</strong> antimicrobial susceptibilities <strong>of</strong> nosocomial pathogens. Effective<br />
recommendation <strong>of</strong> certain antimicrobial agents over others requires access to (<strong>and</strong> financial <strong>and</strong><br />
logistic support for) routine antimicrobial susceptibility testing. Monitoring institutional<br />
resistance patterns is vital in order to make required formulary changes in response to emerging<br />
resistance patterns <strong>and</strong> to determine the most effective agents given prevailing susceptibility<br />
patterns.<br />
Comment<br />
Given the strong association between antibiotic use <strong>and</strong> subsequent infection<br />
(demonstrated for both C. difficile <strong>and</strong> VRE), it is not surprising that changes in antibiotic use<br />
practices can reduce the incidence <strong>of</strong> infection with these 2 pathogens. The majority <strong>of</strong> reviewed<br />
studies demonstrated a significant reduction in the incidence <strong>of</strong> VRE or C. difficile following<br />
interventions to change antibiotic use practice. While these studies all demonstrated short-term<br />
success, future studies should confirm the efficacy <strong>of</strong> such interventions over the long term. In<br />
addition, the effectiveness <strong>and</strong> feasibility <strong>of</strong> combining antibiotic practice strategies with efforts<br />
to enhance barrier precautions (Chapter 13) should be investigated. Finally, the costeffectiveness<br />
<strong>of</strong> such strategies (taking into account both the costs associated with monitoring<br />
<strong>and</strong> maintaining sound antibiotic use practices <strong>and</strong> the costs associated with nosocomial<br />
antibiotic-resistant infections) should be investigated.<br />
144
Table 14.1. Before-after studies <strong>of</strong> practices to improve antibiotic use*<br />
Study Setting <strong>and</strong> Intervention Outcomes Results: before vs. after practice<br />
Elderly care unit <strong>of</strong> a large teaching hospital in<br />
Engl<strong>and</strong>, 1984-85; Changes in empiric<br />
antibiotic regimens 29<br />
Chronic care facility in Baltimore, 1985-86;<br />
multifaceted intervention 21<br />
Veterans Affairs <strong>Medical</strong> Center in Arizona,<br />
1990-92; restriction <strong>of</strong> clindamycin use 28<br />
660-bed Veterans Affairs hospital in<br />
California, 1992-94; removal <strong>of</strong> antibiotic<br />
restrictions 30<br />
703-bed Veterans Affairs <strong>Medical</strong> Center in<br />
Virginia, 1993-94; restriction <strong>of</strong> clindamycin<br />
use 7<br />
557-bed academic medical center in Maryl<strong>and</strong>,<br />
1994; restriction <strong>of</strong> vancomycin use 8<br />
35-bed hematologic malignancy unit in a large<br />
medical center in Engl<strong>and</strong>, 1994-95; sequential<br />
antimicrobial formulary changes 9<br />
Large academic medical center in Virginia,<br />
1994-95; computer-based restriction <strong>of</strong><br />
vancomycin use 10<br />
310-bed Veterans Affairs medical center in<br />
New York, 1995; restriction <strong>of</strong> multiple<br />
antibiotics 27<br />
725-bed teaching hospital in Philadelphia,<br />
1995-96; restriction <strong>of</strong> vancomycin use 11<br />
Level 1 C. difficile infections decreased from 37<br />
to 16 cases (p=0.002).<br />
Level 1 <strong>Patient</strong>s with C. difficile toxin decreased<br />
from 28% to 24% (p=NS); <strong>Patient</strong>s with<br />
C. difficile culture increased from 33%<br />
to 42% (p=NS)<br />
Level 1 C. difficile infections decreased from<br />
7.7 to 1.9 cases/month (p
References<br />
1. Kunin CM, Tupasi T, Craig WA. Use <strong>of</strong> antibiotics. A brief exposition <strong>of</strong> the problem <strong>and</strong><br />
some tentative solutions. Ann Intern Med. 1973;79:555-560.<br />
2. Gaynes R. The impact <strong>of</strong> antimicrobial use on the emergence <strong>of</strong> antimicrobial-resistant<br />
bacteria in hospitals. Infect Dis Clin North Am. 1997;11:757-765.<br />
3. Goldmann DA, Huskins WC. Control <strong>of</strong> nosocomial antimicrobial-resistant bacteria: a<br />
strategic priority for hospitals worldwide. Clin Infect Dis. 1997;24:S139-145.<br />
4. Murray BE. Vancomycin-resistant enterococcal infections. N Engl J Med. 2000;342:710-<br />
721.<br />
5. Johnson S, Gerding DN. Clostridium difficile-associated diarrhea. Clin Infect Dis.<br />
1998;26:1027-1036.<br />
6. Roghmann MC, McCarter RJ, Brewrink J, Cross AS, Morris JG. Clostridium difficile<br />
infection is a risk factor for bacteremia due to vancomycin-resistant enterococci (VRE) in<br />
VRE-colonized patients with acute leukemia. Clin Infect Dis. 1997;25:1056-1059.<br />
7. Climo MW, Israel DS, Wong ES, Williams D, Coudron P, Markowitz SM. Hospital-wide<br />
restriction <strong>of</strong> clindamycin: effect on the incidence <strong>of</strong> Clostridium difficile-associated<br />
diarrhea <strong>and</strong> cost. Ann Intern Med. 1998;128:989-995.<br />
8. Morris JG, Shay DK, Hebden JN, McCarter RJ, Perdue BE, Jarvis W, et al. Enterococci<br />
resistant to multiple antimicrobial agents, including vancomycin. Establishment <strong>of</strong><br />
endemicity in a university medical center. Ann Intern Med. 1995;123:250-259.<br />
9. Bradley SJ, Wilson ALT, Allen MC, Sher HA, Goldstone AH, Scott GM. The control <strong>of</strong><br />
hyperendemic glycopeptide-resistant Enterococcus spp. on a haematology unit by changing<br />
antibiotic usage. J Antimicrob Chemother. 1999;43:261-266.<br />
10. Anglim AM, Klym B, Byers KE, Scheld WM, Farr BM. Effect <strong>of</strong> a vancomycin restriction<br />
policy on ordering practices during an outbreak <strong>of</strong> vancomycin-resistant Enterococcus<br />
faecium. Arch Intern Med. 1997;157:1132-1136.<br />
11. Morgan AS, Brennan PJ, Fishman NO. Impact <strong>of</strong> a vancomycin restriction policy on use<br />
<strong>and</strong> cost <strong>of</strong> vancomycin <strong>and</strong> incidence <strong>of</strong> vancomycin-resistant enterococcus. Ann<br />
Pharmacother. 1997;31:970-973.<br />
12. Stosor V, Peterson LR, Postelnick M, Noskin GA. Enterococcus faecium bacteremia: does<br />
vancomycin resistance make a difference? Arch Intern Med. 1998;158:522-527.<br />
13. Clabots CR, Johnson S, Olson MM, Peterson LR, Gerding DN. Acquisition <strong>of</strong> Clostridium<br />
difficile by hospitalized patients: evidence <strong>of</strong> colonized new admissions as the source <strong>of</strong><br />
infection. J Infect Dis. 1992;166:561-567.<br />
14. Wilcox MH, Smyth ETM. Incidence <strong>and</strong> impact <strong>of</strong> Clostridium difficile infection in the<br />
UK, 1993-1996. J Hosp Infect. 1998;39:181-187.<br />
15. Gerding DN, Olson MM, Peterson LR, et al. Clostridium difficile-associated diarrhea <strong>and</strong><br />
colitis in adults: a prospective case-controlled epidemiologic study. Arch Intern Med.<br />
1986;146:95-100.<br />
16. Yablon SA, Krotenberg, Fruhmann K. Clostridium difficile-related disease: evaluation <strong>and</strong><br />
prevalence among inpatients with diarrhea in two freest<strong>and</strong>ing rehabilitation hospitals.<br />
Arch Phys Med Rehabil. 1993;74:913.<br />
17. Olson MM, Shanholtzer CJ, Lee JT, Gerding DN. Ten years <strong>of</strong> prospective Clostridium<br />
difficile-associated disease surveillance <strong>and</strong> treament at the Minneapolis VA <strong>Medical</strong><br />
Center, 1982-1991. Infect Control Hosp Epidemiol. 1994;15:371-381.<br />
146
18. MacGowan AP, Brown I, Feeney R, Lovering A, McCulloch SY, Reeves DS. Clostridium<br />
difficile associated diarrhea <strong>and</strong> legnth <strong>of</strong> hospital stay. J Hosp Infect. 1995;31:241-244.<br />
19. Riley TV, Codde JP, Rouse IL. Increase length <strong>of</strong> stay due to Clostridium difficile<br />
associated diarrhoea. Lancet. 1995;345:455-456.<br />
20. Kent KC, Rubin MS, Wroblewski L, Hanff PA, Sline W. The impact <strong>of</strong> Clostridium<br />
difficile on a surgical service. Ann Surg. 1998;227:296-301.<br />
21. Bender BS, Laughon BE, Gaydos C, Forman MS, Bennett R, Greenough WB, et al. Is<br />
Clostridium difficile endemic in chronic-care facilties? Lancet. 1986;2:11-13.<br />
22. Spencer RC. Clinical impact <strong>and</strong> associated costs <strong>of</strong> Clostridium difficile-associated<br />
disease. J Antimicrob Chemother. 1998;41(Suppl C):C5-C12.<br />
23. Pestotnik SL, Classen DC, Evans RS. Implementing antibiotic practice guidelines through<br />
computer-assisted decision support: clinical <strong>and</strong> financial outcomes. Ann Intern Med.<br />
1996;124:884-890.<br />
24. Yates RR. New intervention strategies for reducing antibiotic resistance. Chest.<br />
1999;115:24S-27S.<br />
25. Piquette RK. A targeted review <strong>of</strong> vancomycin use. Can J Hosp Pharm. 1991;44:83-7.<br />
26. John JF, Fishman NO. Programmatic role <strong>of</strong> the infectious diseases physician in<br />
controlling antimicrobial costs in the hospital. Clin Infect Dis. 1997;24:471-485.<br />
27. Quale J, L<strong>and</strong>man D, Saurina G, Atwood E, DiTore V, Patel K. Manipulation <strong>of</strong> a hospital<br />
antimicrobial formulary to control an outbreak <strong>of</strong> vancomycin-resistant enterococci. Clin<br />
Infect Dis. 1996;23:1020-1025.<br />
28. Pear SM, Williamson TH, Bettin KM, Gerding DN, Galgiani JN. Decrease in nosocomial<br />
Clostridium difficile-associated diarrhea by restricting clindamycin use. Ann Intern Med.<br />
1994;120:272-277.<br />
29. McNulty C, Logan M, Donald IP, Ennis D, Taylor D, Baldwin RN, et al. Successful<br />
control <strong>of</strong> Clostridium difficile infection in an elderly care unit through use <strong>of</strong> a restrictive<br />
antibiotic policy. J Antimicrob Chemother. 1997;40:707-711.<br />
30. Ho M, Yang D, Wyle FA, Mulligan ME. Increased incidence <strong>of</strong> Clostridium difficileassociated<br />
diarrhea following decreased restriction <strong>of</strong> antibiotic use. Clin Infect Dis.<br />
1996;23 (suppl 1):S102-S106.<br />
31. H<strong>and</strong>werger S, Raucher B, Altarac D, Monka J, Marchione S, Singh KV, et al. Nosocomial<br />
outbreak due to Enterococcus faecium highly resistant to vancomycin, penicillin, <strong>and</strong><br />
gentamicin. Clin Infect Dis. 1993;16:750-755.<br />
32. Chang VT, Nelson K. The role <strong>of</strong> physical proximity in nosocomial diarrhea. Clin Infect<br />
Dis. 2000;31:717-722.<br />
33. Mayfield JL, Leet T, Miller J, Mundy LM. Environmental control to reduce transmission <strong>of</strong><br />
Clostridium difficile. Clin Infect Dis. 2000;31:995-1000.<br />
34. Byers KE, Durbin LJ, Simonton BM, Anglim AM, Adal KA, Farr BM. Disinfection <strong>of</strong><br />
hospital rooms contaminated with vancomycin-resistant Enterococcus faecium. Infect<br />
Control Hosp Epidemiol. 1998;19:261-264.<br />
35. Jacoby GA. Extended-spectrum beta-lactamases <strong>and</strong> other enzymes providing resistance to<br />
oxyimino-beta-lactams. Infect Dis Clin North Am. 1997;11:875-887.<br />
36. Michel M, Gutmann L. Methicillin-resistant Staphylococcus aureus <strong>and</strong> vancomycinresistant<br />
enterococci: therapeutic realities <strong>and</strong> possibilities. Lancet. 1997;349:1901-1906.<br />
37. Smith TL, Pearson ML, Wilcox KR, Cruz C, Lancaster MV, Robinson-Dunn B, et al.<br />
Emergence <strong>of</strong> vancomycin resistance in Staphylococcus aureus. Glycopeptide-Intermediate<br />
Staphylococcus aureus Working Group. N Engl J Med. 1999;340:493-501.<br />
147
38. Archer GL. Staphylococcus aureus: a well-armed pathogen. Clin Infect Dis. 1998;26:1179-<br />
1181.<br />
39. Edmond MB, Wenzel RP, Pasculle AW. Vancomycin-resistant Staphylococcus aureus:<br />
perspectives on measures needed for control. Ann Intern Med. 1996;124:329-334.<br />
148
Chapter 15. Prevention <strong>of</strong> Nosocomial Urinary Tract Infections<br />
Sanjay Saint, MD, MPH<br />
University <strong>of</strong> Michigan School <strong>of</strong> Medicine<br />
Background<br />
Many hospitalized patients require the placement <strong>of</strong> indwelling urinary catheters for days<br />
or even weeks at a time. 1 Only a minority <strong>of</strong> patients develop urinary tract infections because <strong>of</strong><br />
the presence <strong>of</strong> these devices, 2,3 but the frequency <strong>of</strong> their use produces substantial overall<br />
morbidity for patients <strong>and</strong> costs to the health care system. Urinary tract infections (UTIs)<br />
account for up to 40% <strong>of</strong> nosocomial infections, 4,5 with urinary catheter-related infections<br />
causing the vast majority <strong>of</strong> nosocomial UTIs. 6 Each hospital-acquired UTI adds approximately<br />
$675 to the costs <strong>of</strong> hospitalization. When bacteremia develops, this additional cost increases to<br />
at least $2800. 2<br />
Because <strong>of</strong> the substantial complications <strong>and</strong> costs associated with the use <strong>of</strong> urinary<br />
catheters, a number <strong>of</strong> practices have been evaluated in an effort to reduce the incidence <strong>of</strong><br />
urinary catheter-related infections. This chapter reviews the evidence supporting the use <strong>of</strong> silver<br />
alloy coated urinary catheters, <strong>and</strong>, because <strong>of</strong> its similarity, the recently described practice <strong>of</strong><br />
using urinary catheters impregnated with the antibiotic combination <strong>of</strong> minocycline <strong>and</strong><br />
rifampin. Subchapter 15.2 reviews the evidence supporting the use <strong>of</strong> suprapubic catheters as an<br />
alternative to urethral catheters.<br />
Subchapter 15.1. Use <strong>of</strong> Silver Alloy Urinary Catheters<br />
Practice Description<br />
Silver is a highly effective antibacterial substance, which can be applied to various types<br />
<strong>of</strong> catheters. (See Subchapter 16.2 for a discussion <strong>of</strong> intravascular catheters coated with a<br />
combination <strong>of</strong> silver sulfadiazine <strong>and</strong> chlorhexidine). Multiple studies have suggested that<br />
silicone urethral catheters coated with hydrogel <strong>and</strong> silver salts reduce the risk <strong>of</strong> developing<br />
bacteriuria, compared with st<strong>and</strong>ard latex urethral catheters (Foley catheters). As shown in a<br />
recent meta-analysis, this benefit applies to catheters coated with silver alloy (which are coated<br />
on both internal <strong>and</strong> external surfaces <strong>of</strong> the catheter), but not silver oxide (which are coated on<br />
the external catheter surface only). Consequently, this chapter focuses only on studies evaluating<br />
silver alloy catheters, <strong>and</strong> the use <strong>of</strong> catheters coated with antimicrobials. 8<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Almost one million episodes <strong>of</strong> nosocomial UTI occur annually in the United States. 9<br />
Each year approximately 96 million urethral catheters are sold worldwide. Of these, nearly 25%<br />
are sold in the United States. 3 The daily rate <strong>of</strong> bacteriuria in catheterized patients ranges from 3<br />
to 10%, with the incidence directly related to the duration <strong>of</strong> catheterization. 4 Among patients<br />
with bacteriuria, 10 to 25% will develop symptoms <strong>of</strong> local urinary tract infection, 2,10 such as<br />
suprapubic or flank pain. The development <strong>of</strong> catheter-related bacteriuria carries with it a 2.8fold<br />
increased risk <strong>of</strong> death, independent <strong>of</strong> other co-morbid conditions <strong>and</strong> disease severity. 11,12<br />
Bacteremia results from catheter-related bacteriuria in approximately 3% <strong>of</strong> patients, <strong>and</strong><br />
invariably represents a serious complication. 2,3<br />
Beyond the morbidity <strong>and</strong> mortality associated with indwelling catheters, catheter-related<br />
infection results in substantially increased health care costs. Data suggest that each episode <strong>of</strong><br />
149
hospital-acquired symptomatic catheter-related UTI costs an additional $676, <strong>and</strong> each episode<br />
<strong>of</strong> catheter-related nosocomial bacteremia costs a minimum <strong>of</strong> $2836. 2<br />
Estimates from one university hospital, based on data from almost 20 years ago, were<br />
that hospital-acquired UTI led to approximately $204,000 in additional expenses per year. 13<br />
More recent data are unavailable, but the institutional costs attributable to catheter-related<br />
infection are clearly substantial.<br />
Opportunities for Impact<br />
Since catheter-related UTI is the leading cause <strong>of</strong> nosocomial infection in the United<br />
States <strong>and</strong> is associated with increased morbidity <strong>and</strong> costs, any intervention that reduces the<br />
incidence <strong>of</strong> catheter-related UTI is potentially important. Currently, it is unknown what<br />
proportion <strong>of</strong> patients with indwelling catheters receives silver alloy catheters, however it is<br />
likely to be the minority.<br />
Study Designs<br />
As shown in Table 15.1.1, a meta-analysis 7 which included 4 r<strong>and</strong>omized clinical<br />
trials, 14-17 compared the efficacy <strong>of</strong> silver catheters with st<strong>and</strong>ard, non-coated catheters. Five<br />
additional studies 18-22 have appeared since publication <strong>of</strong> this meta-analysis. In 3 <strong>of</strong> these<br />
studies, 18,20,22 the patient represented the unit <strong>of</strong> analysis. Another study employed a r<strong>and</strong>omized<br />
crossover design (Level 1), r<strong>and</strong>omizing wards rather than individual patients. 19 The final study<br />
used a prospective, before-after design at 5 different hospitals (Level 2). 21<br />
The patient populations for these studies included patients on various hospital services<br />
including urology, internal medicine, neurology, <strong>and</strong> the intensive care unit. In general, the<br />
studies included patients expected to be catheterized for at least 2 days. Since the patients resided<br />
in acute care hospitals rather than extended care centers, most were catheterized for 10 days or<br />
less. Several studies specified that patients given concomitant antibiotics were excluded. 15-18<br />
Study Outcomes<br />
The individual trials <strong>and</strong> the meta-analysis focused primarily on the surrogate outcome <strong>of</strong><br />
bacteriuria (Level 2). The definition <strong>of</strong> bacteriuria varied somewhat in the studies. However,<br />
low-level growth from a catheterized specimen (ie, 10 2 colony forming units (CFU) /mL) usually<br />
progresses within days to concentrations <strong>of</strong> greater than 10 4 CFU/mL unless antibiotic therapy is<br />
given. 23 Unfortunately, none <strong>of</strong> the studies was adequately powered to detect a significant<br />
difference in the clinically more important outcomes <strong>of</strong> catheter-related bacteremia or death.<br />
Though bacteriuria is a surrogate endpoint, 24 it is probably appropriate to use since it is a<br />
component <strong>of</strong> the only causal pathway in the disease process between catheterization <strong>and</strong> an<br />
important clinical outcome (eg, symptomatic UTI or catheter-related bacteremia). One study did<br />
report differences in secondary bloodstream infections. 19<br />
150
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The 4 clinical trials 14-17 <strong>of</strong> silver alloy catheters included in the meta-analysis 7 all showed<br />
a significant reduction in the development <strong>of</strong> catheter-associated bacteriuria. As shown in Table<br />
15.1.1, studies published after the meta-analysis have reported more mixed results. Several <strong>of</strong> the<br />
studies have shown a statistically significant benefit <strong>of</strong> silver alloy catheters, but with a smaller<br />
relative risk reduction compared to that reported in the meta-analysis. 19,21,22 However, one study<br />
failed to find a significant benefit associated with silver alloy catheters, 20 <strong>and</strong> another found<br />
benefit from silver alloy catheters in those given such catheters for about 5 days, but not in those<br />
given the catheter for 14 days. 18 A formal update <strong>of</strong> the previous meta-analysis would be helpful,<br />
but is beyond the scope <strong>of</strong> the current report.<br />
Potential for Harm<br />
There is likely minimal harm from the use <strong>of</strong> silver alloy urinary catheters. The one<br />
theoretical harm involves the development <strong>of</strong> antimicrobial resistance. However, since silver is<br />
not used systemically in the form <strong>of</strong> an antimicrobial agent for treatment, the clinical<br />
significance <strong>of</strong> antimicrobial resistance to silver is unclear.<br />
Costs <strong>and</strong> Implementation<br />
Each silver alloy urinary catheter tray costs about $5.30 more than a st<strong>and</strong>ard, non-coated<br />
urinary catheter tray. However, a recent economic evaluation indicates that when all the clinical<br />
<strong>and</strong> economic costs are accounted for, silver alloy urinary catheters may provide both clinical<br />
<strong>and</strong> economic benefits in patients receiving indwelling catheterization for 2 to 10 days. 3 It should<br />
be noted that one <strong>of</strong> the major assumptions made in the economic evaluation is that a certain<br />
proportion <strong>of</strong> patients with bacteriuria develop the clinically important (Level 1) outcomes <strong>of</strong><br />
symptomatic UTI or bacteremia. The economic analysis did not assign any costs to bacteriuria<br />
but did assign costs if patients developed these clinically important outcomes. Additionally,<br />
several <strong>of</strong> the very recent efficacy studies <strong>of</strong> silver alloy catheters 19,21,22 were not included in the<br />
economic analysis. A clinical study, adequately powered to detect both meaningful clinical <strong>and</strong><br />
economic endpoints, would confirm the results <strong>of</strong> this economic evaluation that relied on<br />
modeling techniques. The overall cost <strong>of</strong> universal implementation <strong>of</strong> silver alloy catheters is<br />
unclear.<br />
Comment<br />
The data supporting the use <strong>of</strong> silver alloy urinary catheters to reduce urinary catheterrelated<br />
bacteriuria is reasonably strong. As noted, the incidence <strong>of</strong> bacteriuria, while not<br />
extremely high, carries a high morbidity. It remains unclear whether silver alloy urinary<br />
catheters will also lead to decreases in the clinically more important outcomes <strong>of</strong> catheter-related<br />
bacteremia <strong>and</strong> mortality. Continuing investigation into the impact <strong>of</strong> silver alloy catheters on<br />
these important outcomes <strong>and</strong> their effect on the emergence <strong>of</strong> antibiotic resistance should be<br />
pursued.<br />
151
Of note, catheters coated with antibacterial substances other than silver have also been<br />
evaluated. A recent r<strong>and</strong>omized study 8 found that patients who received antimicrobialimpregnated<br />
catheters coated with minocycline <strong>and</strong> rifampin had significantly lower rates <strong>of</strong><br />
gram-positive bacteriuria than a control group given st<strong>and</strong>ard, non-coated catheters (7.1% vs.<br />
38.2%; p
Table 15.1.1. Studies <strong>of</strong> silver alloy <strong>and</strong> antibiotic-impregnated urethral catheters*<br />
Study Description Design,<br />
Outcomes<br />
Saint,<br />
1998 7<br />
Maki,<br />
1998 22<br />
Verleyen,<br />
1999 18<br />
Bologna,<br />
1999 21<br />
Karchmer,<br />
2000 19<br />
Thibon,<br />
2000 20<br />
Meta-analysis <strong>of</strong> 4 r<strong>and</strong>omized<br />
controlled trials (n=453) <strong>of</strong><br />
silver alloy vs. uncoated urinary<br />
catheters<br />
Prospective, r<strong>and</strong>omized,<br />
double-blind trial <strong>of</strong> silver alloy<br />
(n=407) vs. st<strong>and</strong>ard Foley<br />
(n=443) catheters<br />
Prospective, r<strong>and</strong>omized study<br />
<strong>of</strong> medium-term catheterization<br />
with silver alloy (n=18) vs.<br />
silicone (n=17) catheters after<br />
radical prostatectomy<br />
Prospective, r<strong>and</strong>omized study<br />
<strong>of</strong> short-term catheterization<br />
with silver alloy (n=79) vs. latex<br />
(n=101) catheters<br />
Prospective, blinded study <strong>of</strong><br />
silver alloy vs. st<strong>and</strong>ard latex<br />
Foley catheters in 5 hospitals.<br />
Baseline period ranged from 3-<br />
12 months (mean, 8 months);<br />
intervention period ranged from<br />
7-19 months (mean, 10 months)<br />
12-month r<strong>and</strong>omized crossover<br />
trial <strong>of</strong> catheter-associated<br />
urinary tract infections in<br />
patients with silver-coated <strong>and</strong><br />
uncoated catheters. The ward<br />
was the unit <strong>of</strong> analysis. A cost<br />
analysis was also conducted.<br />
Multicenter, prospective,<br />
r<strong>and</strong>omized, double-blind trial <strong>of</strong><br />
silver alloy (n=90) vs. st<strong>and</strong>ard<br />
(n=109) catheters in patients<br />
Level 1A,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 2,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 2<br />
153<br />
Results: Odds or Risk <strong>of</strong> Bacteriuria†<br />
(unless otherwise noted)<br />
OR 0.24 (95% CI: 0.11-0.52)<br />
RR 0.74 (95% CI: 0.56-0.99)<br />
After 14 days, 50.0% vs. 53.3% (p=NS)<br />
On day 5, 6.3% vs. 11.9% (p
Darouiche,<br />
1999 8<br />
requiring catheterization for >3<br />
days<br />
Multicenter, prospective,<br />
r<strong>and</strong>omized, blinded trial <strong>of</strong><br />
medium-term catheterization<br />
(mean, 14 days) with<br />
minocycline-rifampin<br />
impregnated (n=56) vs. silicone<br />
(n=68) catheters after radical<br />
prostatectomy<br />
Level 1,<br />
Level 2<br />
154<br />
<strong>Patient</strong>s took longer to develop<br />
bacteriuria with antimicrobialimpregnated<br />
catheters than control<br />
catheters (p=0.006 by the log-rank<br />
test)<br />
Overall bacteriuria at day 7: 15.2% vs.<br />
39.7% (p
9. Kunin CM. Urinary Tract Infections: Detection, Prevention, <strong>and</strong> Management. 5th ed.<br />
Baltimore: Williams <strong>and</strong> Wilkins; 1997.<br />
10. Tambyah PA, Maki DG. Catheter-associated urinary tract infection is rarely symptomatic:<br />
a prospective study <strong>of</strong> 1,497 catheterized patients. Arch Intern Med. 2000;160:678-682.<br />
11. Platt R, Polk BF, Murdock B, Rosner B. Mortality associated with nosocomial urinary-tract<br />
infection. N Engl J Med. 1982;307:637-642.<br />
12. Platt R, Polk BF, Murdock B, Rosner B. Reduction <strong>of</strong> mortality associated with<br />
nosocomial urinary tract infection. Lancet. 1983;1:893-897.<br />
13. Krieger JN, Kaiser DL, Wenzel RP. Nosocomial urinary tract infections: secular trends,<br />
treatment <strong>and</strong> economics in a university hospital. J Urol. 1983;130:102-106.<br />
14. Lundeberg T. Prevention <strong>of</strong> catheter-associated urinary-tract infections by use <strong>of</strong> silverimpregnated<br />
catheters [letter]. Lancet. 1986;2:1031.<br />
15. Liedberg H, Lundeberg T. Silver alloy coated catheters reduce catheter-associated<br />
bacteriuria. Br J Urol. 1990;65:379-381.<br />
16. Liedberg H, Lundeberg T, Ekman P. Refinements in the coating <strong>of</strong> urethral catheters<br />
reduces the incidence <strong>of</strong> catheter-associated bacteriuria. An experimental <strong>and</strong> clinical<br />
study. Eur Urol. 1990;17:236-240.<br />
17. Liedberg H, Lundeberg T. Prospective study <strong>of</strong> incidence <strong>of</strong> urinary tract infection in<br />
patients catheterized with bard hydrogel <strong>and</strong> silver-coated catheters or bard hydrogelcoated<br />
catheters [abstract]. J Urol. 1993;149:405A.<br />
18. Verleyen P, De Ridder D, Van Poppel H, Baert L. Clinical application <strong>of</strong> the Bardex IC<br />
Foley catheter. Eur Urol. 1999;36:240-246.<br />
19. Karchmer TB, Giannetta ET, Muto CA, Strain BA, Farr BM. A r<strong>and</strong>omized crossover<br />
study <strong>of</strong> silver-coated urinary catheters in hospitalized patients. Arch Intern Med.<br />
2000;160:3294-3298.<br />
20. Thibon P, Le Coutour X, Leroyer R, Fabry J. R<strong>and</strong>omized multi-centre trial <strong>of</strong> the effects<br />
<strong>of</strong> a catheter coated with hydrogel <strong>and</strong> silver salts on the incidence <strong>of</strong> hospital-acquired<br />
urinary tract infections. J Hosp Infect. 2000;45:117-124.<br />
21. Bologna RA, Tu LM, Polansky M, Fraimow HD, Gordon DA, Whitmore KE.<br />
Hydrogel/silver ion-coated urinary catheter reduces nosocomial urinary tract infection rates<br />
in intensive care unit patients: a multicenter study. Urology. 1999;54:982-987.<br />
22. Maki DG, Knasinski V, Halvorson K, Tambyah PA. A novel silver-hydrogel-impregnated<br />
indwelling urinary catheter reduces CAUTIs: a prospective double-blind trial [abstract].<br />
Infect Control Hosp Epidemiol. 1998;19:682(A10).<br />
23. Stark RP, Maki DG. Bacteriuria in the catheterized patient. What quantitative level <strong>of</strong><br />
bacteriuria is relevant? N Engl J Med. 1984;311:560-564.<br />
24. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann<br />
Intern Med. 1996;125:605-613.<br />
Subchapter 15.2. Use <strong>of</strong> Suprapubic Catheters<br />
Background<br />
As discussed in Subchapter 15.1, the use <strong>of</strong> indwelling urethral catheters results in<br />
substantial morbidity <strong>and</strong> mortality. Given the medical <strong>and</strong> social morbidity associated with<br />
urethral catheters, many clinicians have considered suprapubic catheterization as an alternative<br />
155
to catheterization via the urethra. Suprapubic catheters are inserted in the lower abdomen, an<br />
area with less bacterial colonization than the periurethral region, so that the risk for infection is<br />
thought to be lower than with urethral catheters. Furthermore, although the suprapubic placement<br />
<strong>of</strong> urinary catheters represents a minor surgical procedure, patients may find the result more<br />
comfortable89 <strong>and</strong>, as reviewed below, the development <strong>of</strong> infectious complications is reduced.<br />
Subchapter 15.1 discusses the use <strong>of</strong> silver alloy urinary catheters. The focus <strong>of</strong> this chapter is<br />
the use <strong>of</strong> suprapubic catheters as compared with st<strong>and</strong>ard urethral indwelling catheters in adults.<br />
Practice Description<br />
Suprapubic catheterization typically involves the percutaneous placement <strong>of</strong> a st<strong>and</strong>ard<br />
urinary catheter directly into the bladder. The procedure is performed by urologists using sterile<br />
technique. It is generally performed in the operating room <strong>and</strong> is considered minor surgery.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> Target Problem<br />
In addition to the infectious complications (<strong>and</strong> their associated costs) discussed in<br />
Subchapter 15.1, the use <strong>of</strong> urethral catheters causes substantial patient discomfort. In a recent<br />
study at a Veteran Affairs <strong>Medical</strong> Center, 42% <strong>of</strong> catheterized patients surveyed reported that<br />
the indwelling catheter was uncomfortable, 48% complained that it was painful, <strong>and</strong> 61% noted<br />
that it restricted their activities <strong>of</strong> daily living. 7 Restricted activity reduces patient autonomy <strong>and</strong><br />
may promote other nosocomial complications, such as venous thromboembolism <strong>and</strong> pressure<br />
ulcers. In addition, 30% <strong>of</strong> survey respondents stated that the catheter’s presence was<br />
embarrassing, <strong>and</strong> in unsolicited comments that supplemented the structured questionnaires<br />
several noted that it “hurts like hell.” 7<br />
Opportunities for Impact<br />
Since catheter-related urinary tract infection (UTI) is the leading cause <strong>of</strong> nosocomial<br />
infection in the United States <strong>and</strong> is associated with increased morbidity <strong>and</strong> costs, any<br />
intervention that reduces the incidence <strong>of</strong> catheter-related UTI is potentially important.<br />
Currently, it is unknown what proportion <strong>of</strong> patients who require indwelling urinary catheters<br />
receive suprapubic catheters, however, this practice is uncommon.<br />
Study Design<br />
There have been twelve prospective studies, 8,9,11-17 all but one r<strong>and</strong>omized, 15 comparing<br />
the efficacy <strong>of</strong> suprapubic catheters with st<strong>and</strong>ard, non-coated catheters (Table 15.2.1). In all <strong>of</strong><br />
these studies, the patient was the unit <strong>of</strong> analysis. The patient populations for these studies varied<br />
but generally included patients with acute urinary retention <strong>and</strong> those undergoing various<br />
surgical procedures. Since most <strong>of</strong> the patients evaluated resided in acute care hospitals, the<br />
average duration <strong>of</strong> catheterization was generally less than 14 days.<br />
Study Outcomes<br />
All the trials focused on the outcome <strong>of</strong> bacteriuria. Several <strong>of</strong> the studies also assessed<br />
patient satisfaction <strong>and</strong> the incidence <strong>of</strong> mechanical complications. The definition <strong>of</strong> bacteriuria<br />
varied somewhat in the studies. However, low-level growth from a catheterized specimen (ie,<br />
10 2 colony forming units (CFU)/mL) usually progresses within days to concentrations <strong>of</strong> greater<br />
than 10 4 CFU/mL, unless antibiotic therapy is given. 18 Unfortunately, none <strong>of</strong> the studies was<br />
adequately powered to detect a significant difference in the clinically more important outcomes<br />
<strong>of</strong> catheter-related bacteremia or death. Though bacteriuria is a surrogate endpoint, 19 it is<br />
156
probably appropriate to use since it is a component <strong>of</strong> the only causal pathway in the disease<br />
process between suprapubic catheterization <strong>and</strong> an important clinical outcome (eg, symptomatic<br />
UTI or catheter-related bacteremia).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
As shown in Table 15.2.1, studies comparing suprapubic catheterization with urethral<br />
catheterization have produced mixed results. 8,9,11-17,20-22 Six trials reported lower rates <strong>of</strong><br />
bacteriuria in patients with suprapubic catheters, 11,13,15,16,21,22 <strong>and</strong> 4 trials indicated greater patient<br />
satisfaction with suprapubic as opposed to urethral catheters. 8,13,16,20 In 3 <strong>of</strong> the studies, however,<br />
mechanical complications were higher in those receiving suprapubic catheters. 12,15,16 Of note, 3<br />
studies found that patients given suprapubic catheters have significantly decreased incidence <strong>of</strong><br />
urethral strictures compared with patients who received urethral catheters. 15,23,24 However, the<br />
use <strong>of</strong> prophylactic antibiotics in patients receiving urethral catheters for transurethral resection<br />
<strong>of</strong> the prostate has been shown to significantly reduce the incidence <strong>of</strong> strictures in the anterior<br />
urethra. 25<br />
Potential for Harm<br />
As stated above, the primary problem associated with suprapubic catheter use involves<br />
mechanical complications associated with insertion, most commonly catheter dislodgement or<br />
obstruction, <strong>and</strong> failed introduction. The safe insertion <strong>of</strong> suprapubic indwelling urinary<br />
catheters depends on trained personnel.<br />
Costs <strong>and</strong> Implementation<br />
The cost <strong>of</strong> each suprapubic urinary catheter tray is comparable to the cost <strong>of</strong> each<br />
st<strong>and</strong>ard, non-coated urethral catheter tray. However, the overall initial costs <strong>of</strong> using suprapubic<br />
catheters will no doubt be greater since procedure-related costs are substantially higher for<br />
suprapubic than urethral catheters. Nurses are able to place urethral catheters at the bedside, but<br />
urologists must place suprapubic catheters, <strong>and</strong> the procedure typically occurs in the operating<br />
room. Additionally, it is unclear whether urologists are currently pr<strong>of</strong>icient at the insertion <strong>of</strong><br />
suprapubic catheters given how infrequently they are used. If suprapubic catheters are shown to<br />
be effective, they may have a positive impact on patient care. The cost <strong>of</strong> training individuals in<br />
inserting <strong>and</strong> maintaining the suprapubic catheter is likely to be substantial.<br />
Comment<br />
When compared with st<strong>and</strong>ard urethral indwelling catheters, suprapubic urinary catheters<br />
may reduce urinary catheter-related bacteriuria. Additionally, patient satisfaction may be greater<br />
with suprapubic catheters, although there is also evidence that patients placed with suprapubic<br />
catheters more frequently experience certain mechanical complications. On the other h<strong>and</strong>,<br />
urethral catheters are likely to lead to a higher incidence <strong>of</strong> urethral strictures. Given these mixed<br />
results, conclusions regarding the overall benefit <strong>of</strong> routine suprapubic catheterization cannot<br />
currently be made. However, it would be reasonable to consider conducting a formal metaanalysis<br />
<strong>of</strong> the published trials to answer the question, “Compared with urethral indwelling<br />
catheters, are suprapubic catheters less likely to lead to UTI (as measured by bacteriuria) <strong>and</strong><br />
more likely to lead to enhanced patient satisfaction?” Using explicit inclusion criteria <strong>and</strong><br />
accepted quantitative methods, a meta-analysis 26-28 can <strong>of</strong>ten help clarify the features <strong>of</strong><br />
individual studies that have divergent results. 29 In addition, a possible interaction between gender<br />
<strong>of</strong> the patient <strong>and</strong> type <strong>of</strong> catheter is <strong>of</strong> interest since different pathophysiologic mechanisms<br />
157
underlie the development <strong>of</strong> urethral catheter-related infection in men <strong>and</strong> women. 30 The<br />
possibility <strong>of</strong> adequately evaluating effects within subgroups (eg, those undergoing certain<br />
surgical procedures) because <strong>of</strong> an increased sample size is one <strong>of</strong> the benefits <strong>of</strong> metaanalysis.<br />
31<br />
If formal meta-analysis suggests that suprapubic catheters are less likely to lead to<br />
urinary tract infection <strong>and</strong> more likely to enhance patient satisfaction, at least in some clinical<br />
settings, then these catheters should be considered in the management <strong>of</strong> certain patients. On the<br />
other h<strong>and</strong>, if the meta-analysis finds that urethral catheters are superior to suprapubic catheters,<br />
then use <strong>of</strong> suprapubic catheters, albeit currently quite limited, should be further reduced.<br />
158
Table 15.2.1. Prospective studies comparing suprapubic with urethral catheters<br />
Study Design, <strong>Patient</strong> Bacteriuria (%)† Odds Ratio Comments§<br />
Outcomes Population* Suprapubic Urethral (95% CI)‡<br />
Shapiro,<br />
1982 16<br />
Andersen,<br />
1985 13<br />
Ichsan,<br />
1987 9<br />
Sethia,<br />
1987 11<br />
Schiotz,<br />
1989 12<br />
Horgan,<br />
1992 15<br />
O’Kelley,<br />
1995 8<br />
Ratnaval,<br />
1996 14<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 2,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
General<br />
surgical<br />
patients with<br />
urinary<br />
retention<br />
Women<br />
undergoing<br />
vaginal<br />
surgery<br />
<strong>Patient</strong>s with<br />
acute urinary<br />
retention<br />
General<br />
surgical<br />
patients<br />
requiring<br />
urine output<br />
monitoring<br />
Women<br />
undergoing<br />
vaginal<br />
surgery<br />
Men with<br />
acute urinary<br />
retention due<br />
to prostatic<br />
enlargement<br />
General<br />
surgical<br />
patients<br />
requiring<br />
abdominal<br />
surgery<br />
Men<br />
undergoing<br />
colorectal<br />
surgery<br />
2/25 (8) 21/31<br />
(68)<br />
10/48 (21) 20/44<br />
(45)<br />
3/29 (10) 11/37<br />
(30)<br />
2/32 (6) 16/34<br />
(47)<br />
8/38 (21) 5/40<br />
(12)<br />
10/56 (18) 12/30<br />
(40)<br />
3/28 (11) 3/29<br />
(10)<br />
1/24 (4) 3/26<br />
(12)<br />
159<br />
0.04<br />
(0.01-0.24)<br />
0.32<br />
(0.11-0.86)<br />
0.27<br />
(0.04-1.22)<br />
0.08<br />
(0.01-0.41)<br />
1.87<br />
(0.48-8.01)<br />
0.33<br />
(0.11-0.99)<br />
1.04<br />
(0.13-8.51)<br />
0.33<br />
(0.01-4.60)<br />
Pseudor<strong>and</strong>omized (urethral<br />
catheters used in every third<br />
patient) study; suprapubic<br />
group had less pain but more<br />
mechanical complications<br />
<strong>Patient</strong>s rated acceptability <strong>of</strong><br />
suprapubic catheters greater<br />
None <strong>of</strong> the suprapubic group<br />
complained <strong>of</strong> discomfort<br />
compared with 17 <strong>of</strong> the<br />
patients given urethral<br />
catheters<br />
Decrease in bacteriuria was<br />
more significant in women<br />
than in men<br />
26% <strong>of</strong> suprapubic group<br />
versus 5% <strong>of</strong> urethral group<br />
had mechanical<br />
complications<br />
21% <strong>of</strong> suprapubic group<br />
versus 3% <strong>of</strong> urethral group<br />
had dislodgement; 0% <strong>of</strong><br />
suprapubic group versus 17%<br />
<strong>of</strong> urethral group developed<br />
urethral strictures<br />
Study design unclear, but<br />
probably not r<strong>and</strong>omized;<br />
suprapubic catheters caused<br />
significantly fewer days <strong>of</strong><br />
catheter-related pain<br />
Suprapubic group had fewer<br />
voiding difficulties
Bergman,<br />
1987 21<br />
Abrams,<br />
1980 20<br />
V<strong>and</strong>oni,<br />
1994 22<br />
Perrin,<br />
1997 17<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Women<br />
undergoing<br />
vaginal<br />
surgery for<br />
stress<br />
incontinence<br />
Men with<br />
urinary<br />
retention<br />
<strong>Patient</strong>s<br />
requiring<br />
surgery for<br />
various<br />
indications<br />
<strong>Patient</strong>s<br />
undergoing<br />
rectal surgery<br />
4/24 (17) 17/27<br />
(63)<br />
21/52 (40) 13/50<br />
(26)<br />
0/19 (0) 6/24<br />
(25)<br />
12/49 (24) 29/59<br />
(49)<br />
160<br />
0.26<br />
(0.10-0.68)<br />
1.6<br />
(0.88-2.75)<br />
0<br />
(0-0.95)<br />
0.34<br />
(0.13-0.83)<br />
Length <strong>of</strong> hospital stay was<br />
significantly less (by 1 day)<br />
in the suprapubic catheter<br />
group<br />
12% <strong>of</strong> suprapubic catheter<br />
group found catheter<br />
uncomfortable compared with<br />
64% in the st<strong>and</strong>ard urethral<br />
catheter group (p
6. Saint S, Veenstra DL, Sullivan SD, Chenoweth C, Fendrick AM. The potential clinical <strong>and</strong><br />
economic benefits <strong>of</strong> silver alloy urinary catheters in preventing urinary tract infection.<br />
Arch Intern Med 2000;160:2670-2675.<br />
7. Saint S, Lipsky BA, Baker PD, McDonald LL, Ossenkop K. Urinary catheters: What type<br />
do men <strong>and</strong> their nurses prefer? J Am Geriatr Soc 1999;47:1453-1457.<br />
8. O'Kelly TJ, Mathew A, Ross S, Munro A. Optimum method for urinary drainage in major<br />
abdominal surgery: a prospective r<strong>and</strong>omized trial <strong>of</strong> suprapubic versus urethral<br />
catheterization. Br J Surg 1995;82:1367-1368.<br />
9. Ichsan J, Hunt DR. Suprapubic catheters: a comparison <strong>of</strong> suprapubic versus urethral<br />
catheters in the treatment <strong>of</strong> acute urinary retention. Aust N Z J Surg 1987;57:33-36.<br />
10. Krieger JN, Kaiser DL, Wenzel RP. Nosocomial urinary tract infections: secular trends,<br />
treatment <strong>and</strong> economics in a university hospital. J Urol 1983;130:102-106.<br />
11. Sethia KK, Selkon JB, Berry AR, Turner CM, Kettlewell MG, Gough MH. Prospective<br />
r<strong>and</strong>omized controlled trial <strong>of</strong> urethral versus suprapubic catheterization. Br J Surg<br />
1987;74:624-625.<br />
12. Schiotz HA, Malme PA, Tanbo TG. Urinary tract infections <strong>and</strong> asymptomatic bacteriuria<br />
after vaginal plastic surgery. A comparison <strong>of</strong> suprapubic <strong>and</strong> transurethral catheters. Acta<br />
Obstet Gynecol Sc<strong>and</strong> 1989;68:453-455.<br />
13. Andersen JT, Heisterberg L, Hebjorn S, Petersen K, Stampe Sorensen S, Fischer<br />
Rasmussen W, et al. Suprapubic versus transurethral bladder drainage after<br />
colposuspension/vaginal repair. Acta Obstet Gynecol Sc<strong>and</strong> 1985;64:139-143.<br />
14. Ratnaval CD, Renwick P, Farouk R, Monson JR, Lee PW. Suprapubic versus transurethral<br />
catheterisation <strong>of</strong> males undergoing pelvic colorectal surgery. Int J Colorectal Dis<br />
1996;11:177-179.<br />
15. Horgan AF, Prasad B, Waldron DJ, O'Sullivan DC. Acute urinary retention. Comparison <strong>of</strong><br />
suprapubic <strong>and</strong> urethral catheterisation. Br J Urol 1992;70:149-151.<br />
16. Shapiro J, H<strong>of</strong>fmann J, Jersky J. A comparison <strong>of</strong> suprapubic <strong>and</strong> transurethral drainage for<br />
postoperative urinary retention in general surgical patients. Acta Chir Sc<strong>and</strong> 1982;148:323-<br />
327.<br />
17. Perrin LC, Penfold C, McLeish A. A prospective r<strong>and</strong>omized controlled trial comparing<br />
suprapubic with urethral catheterization in rectal surgery. Aust N Z J Surg 1997;67:554-<br />
556.<br />
18. Stark RP, Maki DG. Bacteriuria in the catheterized patient. What quantitative level <strong>of</strong><br />
bacteriuria is relevant? N Engl J Med 1984;311:560-564.<br />
19. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann<br />
Intern Med 1996;125:605-613.<br />
20. Abrams PH, Shah PJ, Gaches CG, Ashken MH, Green NA. Role <strong>of</strong> suprapubic<br />
catheterization in retention <strong>of</strong> urine. J R Soc Med 1980;73:845-848.<br />
21. Bergman A, Matthews L, Ballard CA, Roy S. Suprapubic versus transurethral bladder<br />
drainage after surgery for stress urinary incontinence. Obstetrics & Gynecology<br />
1987;69:546-549.<br />
22. V<strong>and</strong>oni RE, Lironi A, Tschantz P. Bacteriuria during urinary tract catheterization:<br />
suprapubic versus urethral route: a prospective r<strong>and</strong>omized trial. Acta Chir Belg<br />
1994;94:12-16.<br />
23. Hammarsten J, Lindqvist K, Sunzel H. Urethral strictures following transurethral resection<br />
<strong>of</strong> the prostate. The role <strong>of</strong> the catheter. Br J Urol 1989;63:397-400.<br />
161
24. Hammarsten J, Lindqvist K. Suprapubic catheter following transurethral resection <strong>of</strong> the<br />
prostate: a way to decrease the number <strong>of</strong> urethral strictures <strong>and</strong> improve the outcome <strong>of</strong><br />
operations. J Urol 1992;147:648-651.<br />
25. Hammarsten J, Lindqvist K. Norfloxacin as prophylaxis against urethral strictures<br />
following transurethral resection <strong>of</strong> the prostate: an open, prospective, r<strong>and</strong>omized study. J<br />
Urol 1993;150:1722-1724.<br />
26. Yusuf S. Obtaining medically meaningful answers from an overview <strong>of</strong> r<strong>and</strong>omized<br />
clinical trials. Stat Med 1987;6:281-294.<br />
27. Light RJ. Accumulating evidence from independent studies: what we can win <strong>and</strong> what we<br />
can lose. Stat Med 1987;6:221-231.<br />
28. Hennekens CH, Buring JE, Hebert PR. Implications <strong>of</strong> overviews <strong>of</strong> r<strong>and</strong>omized trials. Stat<br />
Med 1987;6:397-409.<br />
29. Sacks HS, Berrier J, Reitman D, Ancona Berk VA, Chalmers TC. Meta-analyses <strong>of</strong><br />
r<strong>and</strong>omized controlled trials. N Engl J Med 1987;316:450-455.<br />
30. Daifuku R, Stamm WE. Association <strong>of</strong> rectal <strong>and</strong> urethral colonization with urinary tract<br />
infection in patients with indwelling catheters. JAMA 1984;252:2028-2030.<br />
31. Gelber RD, Goldhirsch A. Interpretation <strong>of</strong> results from subset analyses within overviews<br />
<strong>of</strong> r<strong>and</strong>omized clinical trials. Stat Med 1987;6:371-388.<br />
162
Chapter 16. Prevention <strong>of</strong> Intravascular Catheter-Associated Infections<br />
Sanjay Saint, MD, MPH<br />
University <strong>of</strong> Michigan School <strong>of</strong> Medicine<br />
Background<br />
Central venous catheters inserted for short-term use have become common <strong>and</strong> important<br />
devices in caring for hospitalized patients, especially the critically ill. 1 While they have<br />
important advantages (eg, ability to administer large volumes <strong>of</strong> fluid), short-term vascular<br />
catheters are also associated with serious complications, the most common <strong>of</strong> which is infection.<br />
Intravascular catheters are one <strong>of</strong> the most common causes <strong>of</strong> nosocomial bacteremia; 2 <strong>and</strong><br />
catheter-related bloodstream infection (CR-BSI) affects over 200,000 patients per year in the<br />
United States. 3 This chapter focuses primarily on short-term central venous catheters. Two<br />
relatively recent reviews address prevention <strong>of</strong> infection due to other types <strong>of</strong> vascular<br />
catheters. 4,5 We review use <strong>of</strong> maximum barrier precautions (Subchapter 16.1), central venous<br />
catheters coated with antibacterial or antiseptic agents (Subchapter 16.2), <strong>and</strong> use <strong>of</strong><br />
chlorhexidine gluconate at the insertion site (Subchapter 16.3). We review several promising<br />
practices, as well as some common ineffective practices (Subchapter 16.4).<br />
Definitions <strong>and</strong> Microbiology<br />
Catheter-related infections can be subdivided into those that are local <strong>and</strong> those that are<br />
bacteremic. Local infection involves only the insertion site <strong>and</strong> manifests as pericatheter skin<br />
inflammation. Local infection is usually diagnosed when there is evidence <strong>of</strong> an insertion-site<br />
infection (eg, purulence at the exit site). Catheter colonization is defined by growth <strong>of</strong> an<br />
organism from the tip or the subcutaneous segment <strong>of</strong> the removed catheter. Growth <strong>of</strong> greater<br />
than 15 colony-forming units (CFU) using the semiquantitative roll-plate culture technique is<br />
<strong>of</strong>ten used to define catheter colonization. 6 Alternatively, the presence <strong>of</strong> more than 1000 CFUs<br />
per catheter tip segment by quantitative culture using a method such as sonication indicates<br />
evidence <strong>of</strong> catheter colonization. 7 Signs <strong>of</strong> local infection may or may not be present when there<br />
is significant catheter colonization; evidence <strong>of</strong> local infection is observed in at least 5% <strong>of</strong><br />
patients with catheter colonization.<br />
Bacteremic catheter-related infection (<strong>of</strong>ten also referred to as CR-BSI) is defined as a<br />
positive blood culture with clinical or microbiologic evidence that strongly implicates the<br />
catheter as the source <strong>of</strong> infection. 1 This includes: 1) evidence <strong>of</strong> local infection with isolation <strong>of</strong><br />
the same organism from both pus around the site <strong>and</strong> bloodstream; or 2) positive cultures <strong>of</strong> both<br />
the catheter tip (using either semi-quantitative or quantitative methods) <strong>and</strong> bloodstream with the<br />
same organism; or 3) clinical evidence <strong>of</strong> sepsis (eg, fever, altered mental status, hypotension,<br />
leukocytosis) that does not respond to antibiotic therapy, but resolves once the catheter is<br />
removed. 1,5 Some have proposed additional methods <strong>of</strong> diagnosing CR-BSI, including paired<br />
blood cultures (drawn from both the central venous catheter <strong>and</strong> from a noncatheterized vein) 8<br />
<strong>and</strong> a technique in which time to culture positivity for blood drawn from the central venous<br />
catheter is compared with that for the blood drawn from percutaneous venipuncture. 9<br />
The most common organisms causing catheter-related infections are staphylococci, gram<br />
negative rods, <strong>and</strong> C<strong>and</strong>ida species. 10,11 The pathophysiology <strong>of</strong> these infections include several<br />
mechanisms, the most important <strong>of</strong> which involve the skin insertion site <strong>and</strong> the catheter hub. 1<br />
Bacteria migrate from the insertion site on the skin along the external surface <strong>of</strong> the catheter <strong>and</strong><br />
163
then colonize the distal tip. 12,13 The hub can also lead to infection when bacteria are introduced<br />
via the h<strong>and</strong>s <strong>of</strong> medical personnel. These organisms then migrate along the internal surface <strong>of</strong><br />
the lumen <strong>and</strong> may result in bacteremia. 14<br />
Less commonly, catheter-related infection can result from hematogenous seeding <strong>of</strong> the<br />
catheter from another focus 15 or from contaminated infusates. 16<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
A recent quantitative review found that <strong>of</strong> patients in whom st<strong>and</strong>ard, non-coated central<br />
venous catheters are in place on average for 8 days, 25% can be expected to develop catheter<br />
colonization <strong>and</strong> 5% will develop CR-BSI. 17 The risk <strong>of</strong> CR-BSI from this estimate is similar to<br />
the rate reported by the Federal Centers for Disease Control <strong>and</strong> Prevention (CDC). The CDC<br />
has reported an average CR-BSI rate <strong>of</strong> 2.8 to 12.8 infections per 1000 catheter-days for all types<br />
<strong>of</strong> intensive care units <strong>and</strong> average rates <strong>of</strong> 4.5 to 6.1 infections per 1000 catheter-days for<br />
medical/surgical intensive care units. 18<br />
CR-BSI is associated with an increased risk <strong>of</strong> dying, but whether this association is<br />
causal remains controversial. 17 Some argue that hospitalized patients who develop CR-BSI may<br />
differ in their clinical <strong>and</strong> physiologic characteristics, <strong>and</strong> thus may have a higher risk <strong>of</strong> dying<br />
due to intrinsic factors. Proponents <strong>of</strong> this view believe that the development <strong>of</strong> CR-BSI is<br />
primarily a marker <strong>of</strong> severe underlying disease or deficient immunity rather than an<br />
independent risk factor for dying. Unfortunately, the few studies evaluating attributable mortality<br />
due to CR-BSI have conflicting results.<br />
Pittet <strong>and</strong> colleagues estimated that the attributable mortality <strong>of</strong> CR-BSI was 25% in a<br />
matched case-control study. 19,20 Another matched study estimated that the attributable mortality<br />
was 28%. 21 Other investigators have found a much smaller attributable mortality associated with<br />
CR-BSI. DiGiovine et al, in a matched case-control study <strong>of</strong> 136 medical intensive care unit<br />
patients, found a non-significant attributable mortality <strong>of</strong> CR-BSI (4.4%; p=0.51). 22 A recent,<br />
carefully matched cohort study <strong>of</strong> 113 patients by Soufir <strong>and</strong> colleagues also failed to detect a<br />
statistically significant increase in mortality associated with CR-BSI. 23 Nevertheless, given the<br />
small sample size, these authors concluded that their findings are consistent with a 10% to 20%<br />
increased mortality due to CR-BSI. 23 Further research to clarify the mortality associated with<br />
CR-BSI is needed, but the available data are consistent with an attributable mortality <strong>of</strong> CR-BSI<br />
ranging between 4% <strong>and</strong> 20%.<br />
Central venous catheter related infection also leads to increased health care costs. Though<br />
there is substantial variability in the economic estimates, a recent review estimates that an<br />
episode <strong>of</strong> local catheter-related infection leads to an additional cost <strong>of</strong> approximately $400,<br />
while the additional cost <strong>of</strong> CR-BSI ranges from about $6005 to $9738. 17 Some have estimated<br />
that each episode leads to even higher costs, approximately $25,000 per episode. 19,20<br />
164
Prevention<br />
Unnecessarily prolonged catheterization should be avoided. Because <strong>of</strong> the increased risk<br />
<strong>of</strong> infection with prolonged catheterization, many clinicians attempt to reduce this risk with<br />
routine changes <strong>of</strong> the catheter, either over a guidewire or with a new insertion site. However,<br />
the available data do not support this practice. 24 Eyer et al 25 r<strong>and</strong>omized 112 surgical patients<br />
receiving a central venous, pulmonary arterial, or systemic arterial catheter for more than 7 days<br />
into three groups: a) weekly catheter change at a new site; or b) weekly guidewire exchange at<br />
the same site; or c) no routine weekly changes. No significant difference was noted in the<br />
incidence <strong>of</strong> local or bacteremic infection. 25 Cobb <strong>and</strong> colleagues 26 r<strong>and</strong>omized 160 patients with<br />
central venous or pulmonary arterial catheters to either replacement every 3 days at a new site or<br />
over a guidewire, or replacement only when clinically indicated. In those with replacement<br />
catheters at new sites, the risk <strong>of</strong> infectious complications was not decreased <strong>and</strong> the number <strong>of</strong><br />
mechanical complications was increased. Those undergoing routine replacement via a guidewire<br />
exchange showed a trend towards a higher rate <strong>of</strong> bloodstream infections compared with those<br />
who had catheter replacement only when clinically indicated. 26 A recent meta-analysis has<br />
confirmed that routine changes <strong>of</strong> central venous <strong>and</strong> systemic arterial catheters appear<br />
unnecessary; 24 attempts should be made, however, to limit the duration <strong>of</strong> catheterization. Strict<br />
adherence to proper h<strong>and</strong>washing <strong>and</strong> use <strong>of</strong> proven infection control principles is crucial (see<br />
Chapters 13 <strong>and</strong> 14). 27,28<br />
Subchapter 16.1. Use <strong>of</strong> Maximum Barrier Precautions During Central Venous Catheter<br />
Insertion<br />
Practice Description<br />
Catheter-related infections <strong>of</strong>ten result from contamination <strong>of</strong> the central venous catheter<br />
during insertion. Maximum sterile barrier (MSB) precautions may reduce the incidence <strong>of</strong><br />
catheter contamination during insertion <strong>and</strong> thus reduce the rate <strong>of</strong> CR-BSI. MSB precautions<br />
consist <strong>of</strong> the use <strong>of</strong> sterile gloves, long-sleeved gowns, <strong>and</strong> a full-size drape as well as a nonsterile<br />
mask (<strong>and</strong> <strong>of</strong>ten a non-sterile cap) during central venous catheter insertion.<br />
Opportunities for Impact<br />
The proportion <strong>of</strong> patients receiving central venous catheters in whom maximum barrier<br />
precautions are employed is not currently known. If maximum barrier precautions are not used,<br />
then the st<strong>and</strong>ard insertion technique involves the use <strong>of</strong> only sterile gloves <strong>and</strong> a sterile small<br />
drape. Given the additional time required to employ MSB, it is likely that many patients are not<br />
receiving maximum barrier precautions during catheter insertion.<br />
Study Designs<br />
One r<strong>and</strong>omized <strong>and</strong> one non-r<strong>and</strong>omized study have evaluated the use <strong>of</strong> maximum<br />
barrier precautions (Table 16.1.1). The clinical trial r<strong>and</strong>omized 176 patients to catheter insertion<br />
using MSB <strong>and</strong> 167 patients to control (use <strong>of</strong> sterile gloves <strong>and</strong> sterile small drape). 29 A nonr<strong>and</strong>omized<br />
before-after observational evaluation assessed the effect <strong>of</strong> a 1-day course on<br />
infection control practices <strong>and</strong> procedures on physician compliance with MSB use <strong>and</strong> incidence<br />
<strong>of</strong> catheter-infection. 30<br />
165
Study Outcomes<br />
Both studies evaluated rates <strong>of</strong> catheter-related infection (Level 1), including local <strong>and</strong><br />
bloodstream infection.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
There is moderately strong evidence that use <strong>of</strong> maximum barrier precautions decrease<br />
the risk <strong>of</strong> catheter-related infection (Table 16.1.1). Furthermore, the evidence that health care<br />
providers—specifically physicians-in-training—can be taught proper use <strong>of</strong> barrier precautions<br />
<strong>and</strong> thereby decrease the incidence <strong>of</strong> infection is reasonably strong.<br />
Potential for Harm<br />
There is virtually no harm associated with this intervention.<br />
Costs <strong>and</strong> Implementation<br />
The use <strong>of</strong> maximum barrier precautions will cost more than not using this technique in<br />
both materials <strong>and</strong> time. Additionally, teaching health care providers how to properly use<br />
maximum barrier precautions is also time-consuming <strong>and</strong> expensive. Sherertz <strong>and</strong> colleagues<br />
estimated the overall cost <strong>of</strong> their educational program <strong>and</strong> supplies to be $74,081. 30 However,<br />
when the costs <strong>of</strong> preventing catheter-related infection are also included, use <strong>of</strong> MSB has been<br />
estimated to be cost-saving in simplified “back-<strong>of</strong>-the-envelope” cost studies. 29,30 Formal<br />
economic evaluation is required to fully assess the economic consequences <strong>of</strong> full adoption <strong>of</strong><br />
maximum barrier precautions.<br />
Comments<br />
Use <strong>of</strong> MSB appears to be a reasonable method <strong>of</strong> preventing catheter-related infection.<br />
Though achieving full compliance with this method <strong>of</strong> catheter insertion is likely to be<br />
challenging, a relatively simple educational intervention has demonstrated effectiveness in<br />
improving adherence <strong>and</strong> reducing infection rates. Given the excellent benefit-to-harm ratio <strong>of</strong><br />
this patient safety practice, it seems reasonable to strongly consider employing MSB for all<br />
patients requiring central venous catheters. The economic consequences <strong>of</strong> full implementation<br />
<strong>of</strong> this practice are still not entirely clear.<br />
166
Table 16.1.1. Studies <strong>of</strong> vascular catheter-related infection*<br />
Study Description; Intervention Study<br />
Design,<br />
Outcomes<br />
343 patients in a 500-bed cancer referral<br />
center; catheters inserted under maximal<br />
sterile barrier precautions (mask, cap,<br />
sterile gloves, gown, <strong>and</strong> large drape)<br />
vs. control precautions (sterile gloves<br />
<strong>and</strong> small drape only) 29<br />
6 ICUs <strong>and</strong> a step-down unit in an<br />
academic medical center in NC; 1-day<br />
course for physicians-in-training on the<br />
control <strong>of</strong> vascular catheter infection,<br />
emphasizing use <strong>of</strong> full-size sterile<br />
drapes 30<br />
Meta-analysis <strong>of</strong> 12 RCTs (2611<br />
catheters) comparing central venous<br />
catheters coated with<br />
chlorhexidine/silver sulfadiazine with<br />
st<strong>and</strong>ard, non-coated catheters 44<br />
High-risk adult patients at 12 universityaffiliated<br />
hospitals in whom central<br />
venous catheters were expected to<br />
remain in place for 3 days;<br />
minocycline/rifampin vs.<br />
chlorhexidine/silver sulfadiazine<br />
catheters 46<br />
Meta-analysis <strong>of</strong> 7 RCTs (772 catheters)<br />
comparing tunneling with st<strong>and</strong>ard<br />
placement <strong>of</strong> short-term central venous<br />
catheters 61<br />
Meta-analysis <strong>of</strong> 12 RCTs comparing<br />
prophylactic heparin use (in different<br />
forms) with no heparin use on the<br />
following outcomes: central venous<br />
catheter colonization (3 trials), CR-BSI<br />
(4 trials), <strong>and</strong> catheter-related deep<br />
venous thrombosis (7 trials) 59<br />
Level 1,<br />
Level 1<br />
Level 2‡,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
167<br />
Results (p-value or 95% CI)†<br />
CR-BSI per 1000 catheter days: 0.08<br />
vs. 0.5, (p=0.02)<br />
Catheter colonization: 2.3% vs. 7.2%<br />
(p=0.04)<br />
Primary bloodstream infection <strong>and</strong><br />
catheter-related infection decreased<br />
28% (p
Meta-analysis <strong>of</strong> 12 RCTs (918 patients,<br />
1913 catheters) assessing the effect <strong>of</strong><br />
guidewire exchange <strong>and</strong> a prophylactic<br />
replacement strategy (change every 3<br />
days) on central venous catheter-related<br />
colonization (8 trials), exit site infection<br />
(4 trials), bacteremia (8 trials), <strong>and</strong><br />
mechanical complications (9 trials) in<br />
critically ill patients 24<br />
Level 1A,<br />
Level 1<br />
168<br />
Catheter colonization: RR 1.26 (0.87-<br />
1.84)<br />
Exit site infection: RR 1.52 (0.34-6.73)<br />
Bacteremia: RR 1.72 (0.89-3.33)<br />
Mechanical complications: RR 0.48<br />
(0.12-1.91)<br />
Prophylactic catheter replacement<br />
every 3 days was not found to be<br />
better than as-needed replacement<br />
* CI indicates confidence interval; CR-BSI, catheter-related bloodstream infection; OR, odds<br />
ratio; RCT, r<strong>and</strong>omized controlled trial; <strong>and</strong> RR, relative risk.<br />
† Results are reported as intervention group vs. control (st<strong>and</strong>ard or usual care) group.<br />
‡ Prospective before-after study design.
Subchapter 16.2. Use <strong>of</strong> Central Venous Catheters Coated with Antibacterial or<br />
Antiseptic Agents<br />
Practice Description<br />
Recent studies have indicated that central venous catheters coated with antimicrobial<br />
agents reduce the incidence <strong>of</strong> catheter-related bloodstream infection (CR-BSI). Implementing<br />
use <strong>of</strong> these catheters would be simple, primarily involving the replacement <strong>of</strong> st<strong>and</strong>ard, noncoated<br />
vascular catheters. However, these catheters, such as chlorhexidine/silver sulfadiazineimpregnated<br />
catheters <strong>and</strong> minocycline/rifampin-coated catheters, are more expensive than<br />
st<strong>and</strong>ard catheters. Thus, the cost-effectiveness <strong>of</strong> these catheters needs to be considered by<br />
decision makers.<br />
Opportunities for Impact<br />
Currently, it is not known precisely what proportion <strong>of</strong> patient who require central<br />
venous catheterization receive an antimicrobial catheter, however, it is probably the minority <strong>of</strong><br />
patients.<br />
Study Designs<br />
Multiple r<strong>and</strong>omized trials have compared chlorhexidine/silver sulfadiazine central<br />
venous catheters with st<strong>and</strong>ard, non-coated central venous catheters. 31-43 In addition, a recent<br />
meta-analysis used a fixed effects model to combine the results <strong>of</strong> these chlorhexidine/silver<br />
sulfadiazine trials. 44 A large, multicenter study has compared minocycline/rifampin coated<br />
catheters with non-coated, st<strong>and</strong>ard catheters. 45 Additionally, a recent multicenter r<strong>and</strong>omized<br />
trial <strong>of</strong> minocycline/rifampin versus chlorhexidine/silver sulfadiazine catheters has also been<br />
reported. 46 The majority <strong>of</strong> the patients enrolled in the individual studies cited above had a<br />
central venous catheter in place for 8 days on average (range <strong>of</strong> average duration, 5 to 11 days).<br />
Details <strong>of</strong> the characteristics <strong>and</strong> results <strong>of</strong> the trials comparing central venous catheters coated<br />
with chlorhexidine/silver sulfadiazine to control catheters are in Tables 16.2.1 <strong>and</strong> 16.2.2.<br />
Study Outcomes<br />
Most studies reported the incidence <strong>of</strong> catheter colonization <strong>and</strong> CR-BSI. Though the<br />
precise outcome definitions in some <strong>of</strong> the studies varied, in general the definition <strong>of</strong> catheter<br />
colonization <strong>and</strong> CR-BSI used in most <strong>of</strong> these studies was explicit <strong>and</strong> appropriate.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The evidence for the efficacy <strong>of</strong> chlorhexidine/silver sulfadiazine catheters is fairly<br />
substantial. The recent meta-analysis found a statistically significant decrease in the incidence <strong>of</strong><br />
CR-BSI (odds ratio 0.56, 95% CI: 0.37-0.84). 44 There is also reasonable evidence that<br />
minocycline-rifampin catheters reduce the risk <strong>of</strong> CR-BSI compared with st<strong>and</strong>ard, non-coated<br />
catheters. The recent r<strong>and</strong>omized trial <strong>of</strong> minocycline/rifampin versus chlorhexidine/silver<br />
sulfadiazine catheters found a significant <strong>and</strong> clinically important decrease in the incidence <strong>of</strong><br />
CR-BSI in the group <strong>of</strong> patients using minocycline/rifampin compared with chlorhexidine/silver<br />
sulfadiazine catheters (0.3% vs. 3.4%, p
concern. Although there have been no reports <strong>of</strong> hypersensitivity reactions to<br />
chlorhexidine/silver sulfadiazine impregnated central venous catheters in the United States (out<br />
<strong>of</strong> more than 2.5 million sold), 13 cases <strong>of</strong> immediate hypersensitivity reactions have been<br />
reported in Japan, including one potentially associated death. 47 There were 117,000 antisepticimpregnated<br />
catheters sold in Japan before their use was halted because <strong>of</strong> these cases. 47 It is not<br />
clear why there have been no reports <strong>of</strong> hypersensitivity reactions in the U.S; this heterogeneity<br />
may be caused by a higher previous exposure <strong>of</strong> patients in Japan to chlorhexidine or by a<br />
genetic predisposition.<br />
Minocycline <strong>and</strong> rifampin are both occasionally used as systemic antimicrobial agents;<br />
thus, their use on catheters raises the important theoretical issue <strong>of</strong> increased antimicrobial<br />
resistance. At this time, there has been no conclusive evidence that antimicrobial resistance has<br />
or will increase due to the use <strong>of</strong> these catheters.<br />
Costs <strong>and</strong> Implementation<br />
Formal <strong>and</strong> informal economic analyses indicate that central venous catheters coated with<br />
antibacterial agents (such as chlorhexidine/silver sulfadiazine or minocycline/rifampin) are likely<br />
to lead to both clinical <strong>and</strong> economic advantages in selected patients. In terms <strong>of</strong> formal<br />
economic comparisons, a recent analysis compared chlorhexidine/silver sulfadiazine catheters to<br />
st<strong>and</strong>ard catheters <strong>and</strong> found that chlorhexidine/silver sulfadiazine catheters lead to both clinical<br />
<strong>and</strong> economic advantages in patients receiving central venous catheterization for 2 to 10 days<br />
<strong>and</strong> who were considered high risk for infection (ie, critically ill or immunocompromised<br />
patients). Specifically, the chlorhexidine/silver sulfadiazine catheters led to a significant decrease<br />
in the incidence <strong>of</strong> CR-BSI <strong>and</strong> death, <strong>and</strong> a cost savings <strong>of</strong> approximately $200 per catheter<br />
used. 47 Importantly, the risk <strong>of</strong> hypersensitivity reaction to the chlorhexidine/silver sulfadiazine<br />
catheters was considered in the analysis, but had little effect on the overall clinical <strong>and</strong> economic<br />
outcomes. 47<br />
However, given the recently demonstrated efficacy <strong>of</strong> the minocycline/rifampin catheter<br />
compared with the chlorhexidine/silver sulfadiazine catheter, 46 a formal cost-effectiveness<br />
analysis comparing these two types <strong>of</strong> coated catheters is necessary. This is especially important<br />
since the minocycline/rifampin catheter costs about $9 more per catheter than the<br />
chlorhexidine/silver sulfadiazine catheter.<br />
Implementation <strong>of</strong> either <strong>of</strong> these catheters would be straightforward. Stocking the<br />
appropriate antimicrobial catheter in areas <strong>of</strong> the hospital that are likely to require such catheters<br />
(eg, intensive care unit, operative room, hematology-oncology floor) would be a relatively<br />
simple way <strong>of</strong> translating the research findings into actual practice.<br />
170
Comment<br />
In light <strong>of</strong> the substantial clinical <strong>and</strong> economic burden <strong>of</strong> catheter-related infection,<br />
hospital personnel should adopt proven cost-effective methods to reduce this common <strong>and</strong><br />
important nosocomial complication. The bulk <strong>of</strong> the evidence supports the use <strong>of</strong> either<br />
chlorhexidine/silver sulfadiazine or minocycline/rifampin central venous catheters rather than<br />
st<strong>and</strong>ard (non-coated) catheters in high-risk patients requiring short-term central venous<br />
catheterization (eg, for 2 to 10 days). Choosing between the 2 antimicrobial catheters requires a<br />
formal cost-effectiveness analysis since the minocycline/rifampin catheter costs significantly<br />
more than the chlorhexidine/silver sulfadiazine catheter. There are 2 primary issues that should<br />
be addressed when comparing these catheters: the expected duration <strong>of</strong> catheterization <strong>and</strong> the<br />
risk <strong>of</strong> antibiotic resistance to the patient, the hospital, <strong>and</strong> society. Though each<br />
minocycline/rifampin catheter costs more than the chlorhexidine/silver sulfadiazine catheter,<br />
using minocycline/rifampin catheters may actually result in cost-savings for at least some patient<br />
populations given their improved overall efficacy. Of note, the improved efficacy <strong>of</strong> the<br />
minocycline/rifampin catheters may be a result <strong>of</strong> coating both the internal <strong>and</strong> external surfaces<br />
with these substances; the chlorhexidine/silver sulfadiazine catheters evaluated to date have only<br />
had the external surface coated with the antiseptic combination.<br />
171
Table 16.2.1. Characteristics <strong>of</strong> trials comparing central venous catheters coated with<br />
chlorhexidine/silver sulfadiazine to control catheters*<br />
Study Description Number <strong>of</strong><br />
Catheters<br />
(Treatment,<br />
Control)<br />
Tennenberg 31 : 282 hospital patients<br />
(137 treatment, 145 control) in<br />
variety <strong>of</strong> settings; double- <strong>and</strong><br />
triple-lumen catheters without<br />
exchanges over guidewires<br />
Maki 32 : 158 ICU patients (72<br />
treatment, 86 control); triple-lumen<br />
catheters with catheter exchanges<br />
over guidewires<br />
van Heerden 33 §: 54 ICU patients (28<br />
treatment, 26 control); triple-lumen<br />
catheters without catheter exchanges<br />
over guidewires<br />
Hannan 34 : ICU patients; triple-lumen<br />
catheters<br />
Bach 35 §: 26 ICU patients (14<br />
treatment, 12 control); triple-lumen<br />
catheters without catheter exchanges<br />
over guidewires<br />
Bach 36 §: 133 surgical patients (116<br />
treatment, 117 control); double- <strong>and</strong><br />
triple-lumen cathetes without<br />
exchanges over guidewires<br />
Heard 37 §: 111 SICU patients (107<br />
treatment, 104 control); triple-lumen<br />
catheters with exchanges over<br />
guidewires<br />
Collin 38 : 119 ER/ICU patients (58<br />
treatment, 61 control); single-,<br />
double-, <strong>and</strong> triple-lumen catheters<br />
with exchanges over guidewires<br />
Ciresi 39 §: 191 patients receiving<br />
TPN (92 treatment, 99 control);<br />
triple-lumen catheters with<br />
exchanges over guidewires<br />
Pemberton 40 : 72 patients receiving<br />
TPN (32 treatment, 40 control);<br />
triple-lumen catheters without<br />
exchanges over guidewires<br />
137,<br />
145<br />
208,<br />
195<br />
28,<br />
26<br />
68,<br />
60<br />
14,<br />
12<br />
116,<br />
117<br />
151,<br />
157<br />
98,<br />
139<br />
124,<br />
127<br />
32,<br />
40<br />
172<br />
Mean Catheter<br />
Duration in<br />
Days<br />
(Treatment,<br />
Control)<br />
5.1,<br />
53<br />
6.0,<br />
6.0<br />
6.6,<br />
6.8<br />
7,<br />
8<br />
7.0,<br />
7.0<br />
7.7,<br />
7.7<br />
8.5,<br />
9<br />
9.0,<br />
7.3<br />
9.6,<br />
9.1<br />
10,<br />
11<br />
Catheter<br />
Colonization†<br />
SQ (IV, SC,<br />
>15 CFU)<br />
SQ (IV, >15<br />
CFU)<br />
SQ (IV, >15<br />
CFU)<br />
SQ (IV, >10 3<br />
CFU) <br />
QN (IV, >10 3<br />
CFU)<br />
QN (IV, >10 3<br />
CFU)<br />
SQ (IV, SC,<br />
>14 CFU)<br />
SQ (IV, SC,<br />
>15 CFU)<br />
SQ (IV, SC,<br />
>15 CFU)<br />
Catheter-<br />
Related<br />
Bloodstream<br />
Infection†<br />
SO (IV, SC,<br />
site), CS, NS<br />
SO (>15 CFU,<br />
IV, hub, inf)‡<br />
NR<br />
SO (IV, >10 3<br />
CFU), NS<br />
NR<br />
SO (IV)<br />
SO (IV, SC,<br />
>4 CFU)<br />
SO (IV, SC)<br />
SO (IV, SC)<br />
NR SO (IV), Res,<br />
NS
Ramsay 41 §: 397 hospital patients<br />
(199 treatment, 189 control) in a<br />
variety <strong>of</strong> settings; triple-lumen<br />
catheters without exchanges over<br />
guidewires<br />
Trazzera 42 §: 181 ICU/BMT patients<br />
(99 treatment, 82 control); triplelumen<br />
catheters with exchanges over<br />
guidewires<br />
George 43 : Transplant patients; triplelumen<br />
catheters without exchanges<br />
over guidewires<br />
199,<br />
189<br />
123,<br />
99<br />
44,<br />
35<br />
173<br />
10.9,<br />
10.9<br />
11.2,<br />
6.7<br />
SQ (IV, SC,<br />
>15 CFU)<br />
SQ (IV, >15<br />
CFU)<br />
NR SQ (IV, >5<br />
CFU)<br />
SO (IV, SC)<br />
SO (IV, >15<br />
CFU)<br />
SO (IV)<br />
* BMT indicates bone marrow transplant; CFU, colony forming units; CS, clinical signs <strong>of</strong><br />
systemic infection; ER, emergency room; ICU, intensive care unit; IV, intravascular catheter<br />
segment; inf, catheter infusate; NR, not reported; NS, no other sources <strong>of</strong> infection; QN,<br />
quantitative culture; Res, resolution <strong>of</strong> symptoms upon catheter removal; SC, subcutaneous<br />
catheter segment; SICU, surgical intensive care unit; site, catheter insertion site; SO, same<br />
organism isolated from blood <strong>and</strong> catheter; SQ, semi-quantitative culture; <strong>and</strong> TPN, total<br />
parenteral nutrition.<br />
† Catheter segments (or site) cultured <strong>and</strong> criteria for a positive culture are given in parenthesis<br />
‡ Organism identity confirmed by restriction-fragment subtyping<br />
§ Additional information provided by author (personal communications, 1/98-3/98)<br />
Culture method reported as semiquantitative; criteria for culture growth suggests quantitative<br />
method
Study<br />
Table 16.2.2. Results <strong>of</strong> trials comparing central venous catheters coated with<br />
chlorhexidine/silver sulfadiazine to control catheters*<br />
Tennenberg 31<br />
Maki 32<br />
Catheter Colonization Catheter-related Bloodstream Infection<br />
No. (%) Positive Odds Ratio<br />
(95% CI)<br />
No. (%) Positive<br />
Treatment Control<br />
Treatment Control<br />
174<br />
Odds Ratio<br />
(95% CI)<br />
8 (5.8%) 32 (22.1%) 0.22 (0.10-0.49) 5 (3.6%) 9 (6.2%) 0.57 (0.19-1.75)<br />
28 (13.5%) 47 (24.1%) 0.49 (0.29-0.82) 2 (1.0%) 9 (4.6%) 0.20 (0.04-0.94)<br />
van Heerden 33 † 4 (14.3%) 10 (38.5%) 0.27 (0.07-1.00) – – –<br />
Hannan 34<br />
22 (32.4%) 22 (36.7%) 0.83 (0.40-1.72) 5 (7.4%) 7 (11.7%) 0.60 (0.18-2.00)<br />
Bach 35 † 0 (0%) 4 (33.3%) 0 (0-0.65) – – –<br />
Bach 36 † 2 (1.7%) 16 (13.7%) 0.11 (0.02-0.49) 0 (0%) 3 (2.6%) 0 (0-1.28)<br />
Heard 37 † 60 (39.7%) 82 (52.2%) 0.60 (0.38-0.95) 5 (3.3%) 6 (3.8%) 0.86 (0.26-2.89)<br />
Collin 38<br />
2 (2.0%) 25 (18.0%) 0.10 (0.02-0.41) 1 (1.0%) 4 (2.9%) 0.35 (0.04-3.16)<br />
Ciresi 39 † 15 (12.1%) 21(16.5%) 0.69 (0.34-1.42) 13 (10.5%) 14 (11.0%) 0.95 (0.43-2.10)<br />
Pemberton 40<br />
– – – 2 (6.3%) 3 (7.5%) 0.82 (0.13-5.24)<br />
Ramsay 41 † 45 (22.6%) 63 (33.3%) 0.58 (0.37-0.92) 1 (0.5%) 4 (2.1%) 0.23 (0.03-2.11)<br />
Trazzera 42 † 16 (13.0%) 24 (24.2%) 0.47 (0.23-0.94) 4 (3.3%) 5 (5.1%) 0.63 (0.17-2.42)<br />
George 43<br />
10 (22.7%) 25 (71.4%) 0.12 (0.04-0.33) 1 (2.3%) 3 (8.6%) 0.25 (0.02-2.50)<br />
* CI indicates confidence interval.<br />
† Additional information provided by author (personal communications, 1/98-3/98)
Subchapter 16.3. Use <strong>of</strong> Chlorhexidine Gluconate at the Central Venous Catheter<br />
Insertion Site<br />
Practice Description<br />
Microbial populations on the skin are routinely suppressed with antiseptic agents prior to<br />
catheter insertion. Using an antiseptic solution for skin disinfection at the catheter insertion site<br />
helps to prevent catheter-related infections. The physician uses an agent that has antimicrobial<br />
properties to thoroughly cleanse the skin just prior to insertion <strong>of</strong> a central venous catheter. In<br />
the United States, povidone-iodine (PI) is overwhelmingly the most commonly used agent for<br />
this purpose. Recently, several studies have compared the efficacy <strong>of</strong> PI <strong>and</strong> chlorhexidine<br />
gluconate (CHG) solutions in reducing vascular catheter-related infections.<br />
Opportunities for Impact<br />
If PI is the most commonly used agent for site disinfection in the United States even<br />
though CHG may be superior, substantial opportunity exists for impact by switching to CHG.<br />
Study Designs<br />
The study characteristics <strong>of</strong> 6 r<strong>and</strong>omized trials 48-53 comparing any type <strong>of</strong> CHG solution<br />
with PI solution for vascular catheter site care are shown in Table 16.3.1. The mean duration <strong>of</strong><br />
catheterization for the CHG <strong>and</strong> PI groups was comparable in most <strong>of</strong> the studies. There was no<br />
significant difference in the sites at which catheters were inserted between the CHG <strong>and</strong> PI<br />
groups. Several formulations <strong>of</strong> CHG were used, including 9,12-14 an alcoholic solution <strong>and</strong> an<br />
aqueous solution. All studies used 10% PI solution for the control arm.<br />
Study Outcomes<br />
All studies 48-53 evaluated catheter colonization (Level 2 outcome) <strong>and</strong> all but one 52<br />
evaluated CR-BSI (Level 1 outcome). All studies evaluating CR-BSI as an outcome required the<br />
recovery <strong>of</strong> the same microbial species from both the catheter segment <strong>and</strong> a blood culture.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Most clinical trials have revealed that the use <strong>of</strong> CHG solution results in a significant<br />
decrease in catheter colonization, but the evidence is not clear for CR-BSI (Table 16.3.2). Most<br />
<strong>of</strong> the individual trials showed a trend in reducing CR-BSI incidence in patients using CHG<br />
solution. The lack <strong>of</strong> significant results may be a result <strong>of</strong> insufficient statistical power in the<br />
individual studies. A formal meta-analysis <strong>of</strong> the published trials would be valuable in assessing<br />
the comparative efficacy <strong>of</strong> PI versus CHG for central venous catheter site disinfection. Using<br />
explicit inclusion criteria <strong>and</strong> accepted quantitative methods, a meta-analysis 54-56 can <strong>of</strong>ten help<br />
clarify the features <strong>of</strong> individual studies that have divergent results 57 <strong>and</strong> increase statistical<br />
power since several small studies can be pooled. 58<br />
Potential for Harm<br />
Only one study reported adverse effects from the use <strong>of</strong> either antiseptic solution. Maki et<br />
al 48 found erythema at the insertion site in 28.3% <strong>of</strong> catheters in the PI group <strong>and</strong> in 45.3% <strong>of</strong><br />
catheters in the CHG group (p=0.0002). However, there was no statistically significant<br />
difference in erythema among these 2 groups <strong>and</strong> those patients whose site was disinfected with<br />
alcohol. Hypersensitivity reactions to chlorhexidine-silver sulfadiazine impregnated central<br />
175
venous catheters <strong>and</strong> to use <strong>of</strong> CHG for bathing have been reported. Hypersenstivity reactions<br />
were not reported in any <strong>of</strong> the studies, but clinicians should be aware <strong>of</strong> such potential side<br />
effects. Another concern is the development <strong>of</strong> bacterial resistance. However, there have been<br />
few reports <strong>of</strong> bacterial resistance to CHG despite its widespread use for several decades.<br />
Costs <strong>and</strong> Implementation<br />
The cost <strong>of</strong> CHG is approximately twice that <strong>of</strong> PI with an absolute difference <strong>of</strong> $0.51<br />
(approximately $0.92 versus $0.41 for a quantity sufficient to prepare a central venous catheter<br />
insertion site). If meta-analysis suggests that CHG use is effective in reducing the risk <strong>of</strong> CR-<br />
BSI, a formal economic evaluation <strong>of</strong> this issue is required.<br />
Comment<br />
The use <strong>of</strong> chlorhexidine gluconate rather than povidone-iodine solution for catheter site<br />
care may be an effective <strong>and</strong> simple measure for improving patient safety by reducing vascular<br />
catheter-related infections. Formal meta-analysis <strong>and</strong> economic evaluations are required before<br />
strongly recommending that CHG replace PI for central venous catheter site disinfection in<br />
appropriate patient populations.<br />
176
Table 16.3.1. Characteristics <strong>of</strong> studies comparing chlorhexidine gluconate (CHG) <strong>and</strong><br />
povidone-iodine (PI) solutions for vascular catheter site care*<br />
Study Description† Number <strong>of</strong><br />
Catheters<br />
(Treatment,<br />
Control)<br />
Maki 48 : 441 ICU<br />
patients (2% aqueous<br />
CHG solution in 214, PI<br />
in 227)<br />
Sheehan 49 : 189 ICU<br />
patients (2% aqueous<br />
CHG solution in 94, PI<br />
in 95)<br />
Meffre 50 : 1117 hospital<br />
patients (CHG solution<br />
<strong>of</strong> 0.5% alcohol 70% in<br />
568, PI in 549)<br />
Mimoz 51 : ICU patients<br />
(Biseptine®§ vs. PI)<br />
Cobett <strong>and</strong> LeBlanc 52 :<br />
244 hospital patients<br />
(0.5% alcohol 70% in 8,<br />
PI in 161)<br />
Humar et al 53 : 3374 ICU<br />
patients (0.5% alcohol<br />
in 193 <strong>and</strong> 181/193<br />
Mean<br />
Catheter<br />
Duration in<br />
Days<br />
(Treatment,<br />
Control)<br />
177<br />
Catheter<br />
Colonization‡<br />
Catheter-<br />
Related<br />
Bloodstream<br />
Infection‡<br />
214, 227 5.3, 5.3 SQ (>15 CFU) CX, NoSource,<br />
Sx<br />
169,177 NA SQ (>15 CFU) CX, NoSource,<br />
Sx<br />
568, 549 1.6, 1.6 SQ (>15 CFU) or<br />
QN (>10 3<br />
CFU/mL)<br />
170, 145 4.5, 3.9 QN (>10 3<br />
XFU/mL)<br />
83, 161 1.6, 1.7 SQ (>15 CFU) NA<br />
[Local or Sx] or<br />
[CX, NoSource]<br />
CX, Sx<br />
193, 181 5.3, 6. SQ (>15 CFU) CX, Molec,<br />
NoSource<br />
* CFU indicates colony forming units; CX, same organism or species matched between blood<br />
<strong>and</strong> catheter segment culture; ICU: intensive care units; Local: local signs <strong>of</strong> infection; Molec:<br />
same organism confirmed by molecular subtyping; NA:not available; NoSource: no other<br />
source <strong>of</strong> infection; QN: quantitative; Sx: clinical symptoms <strong>of</strong> bloodstream infection; SQ:<br />
semiquantitative.<br />
† All studies used 10% povidone-iodine solution.<br />
‡ Catheter segments (or site) cultured <strong>and</strong> criteria for a positive culture are given in parenthesis.<br />
§ Biseptine® consists <strong>of</strong> 0.25% chlorhexidine gluconate, 0.025% benzalkonium chloride, 4%<br />
benzyl alcohol.<br />
Required one <strong>of</strong> the following symptoms: fever, erythema, heat at the site, <strong>and</strong> pain.
Table 16.3.2. Results <strong>of</strong> Studies Comparing Chlorhexidine Gluconate (CHG) <strong>and</strong> Povidoneiodine<br />
(PI) Solutions for Vascular Catheter Site Care *<br />
Maki 48<br />
Sheehan 9<br />
Meffre 50<br />
Mimoz 51<br />
Cobett <strong>and</strong><br />
LeBlanc 52 †<br />
Humar 53<br />
Catheter Colonization<br />
(Positive Cultures)<br />
CHG PI Solution<br />
Solution<br />
RR (95% CI)<br />
CHG vs. PI<br />
178<br />
Catheter Related<br />
Bloodstream Infection<br />
CHG PI Solution<br />
Solution<br />
RR (95% CI)<br />
CHG vs. PI<br />
5/214 21/227 0.25 (0.10,0.66) 1/214 6/227 0.18 (0.02,1.46)<br />
3/169 12/177 0.22 (0.06,0.75) 1/169 1/177 1.05 (0.07,16.61)<br />
9/568 22/549 0.40 (0.18,0.85) 3/568 3/549 0.97 (0.20,4.77)<br />
12/170 24/145 0.43 (0.22,0.82) 3/170 4/145 0.64 (0.15,2.81)<br />
6/83 23/161 0.49 (0.31,0.77) - - -<br />
36/116 27/116 1.33 (0.87,2.04) 4/193 5/181 0.75 (0.20,2.75)<br />
* CI indicates confidence interval; RR, relative risk.<br />
† Additional information was provided by authors<br />
Subchapter 16.4. Other <strong>Practices</strong><br />
<strong>Practices</strong> That Appear Promising<br />
Use <strong>of</strong> heparin with central venous catheters. Because an association has been shown between<br />
thrombus formation <strong>and</strong> catheter-related infection, clinicians usually use heparin, in a variety <strong>of</strong><br />
forms: 1) as flushes to fill the catheter lumens between use; 2) injected subcutaneously; or 3)<br />
bonded on the catheter. A meta-analysis <strong>of</strong> 12 r<strong>and</strong>omized trials evaluating prophylactic use <strong>of</strong><br />
heparin in patients using central venous catheters has shown that prophylactic heparin decreases<br />
catheter-related venous thrombosis (Level 2 outcome; RR 0.43, 95% CI: 0.23-078) <strong>and</strong> bacterial<br />
colonization (Level 2 outcome; RR 0.18, 95% CI: 0.06-0.60) <strong>and</strong> may decrease CR-BSI (Level 1<br />
outcome; RR 0.26, 95% CI: 0.07-1.03). 59 Since subcutaneous heparin also <strong>of</strong>fers benefit in<br />
reducing venous thromboembolism in certain patient populations (see Chapter 31), this is likely<br />
to be a reasonable strategy even though CR-BSIs have not definitely been shown to be reduced.<br />
However use <strong>of</strong> heparin is associated with several side effects, such as heparin-induced<br />
thrombocytopenia <strong>and</strong> bleeding.<br />
Tunneling short-term central venous catheters. Since the primary site <strong>of</strong> entry for<br />
microorganisms on the central venous catheter is the site <strong>of</strong> cutaneous insertion, 60 tunneling the<br />
catheter through the subcutaneous tissue may decease the incidence <strong>of</strong> infection. Several trials<br />
have evaluated the effect <strong>of</strong> tunneling on catheter-related infection. A recent meta-analysis has<br />
summarized the potential benefit. 61 The meta-analysis included 7 trials <strong>and</strong> found that compared<br />
with patients receiving st<strong>and</strong>ard catheter placement, tunneling decreased bacterial colonization<br />
(Level 2 outcome; RR 0.61, 95% CI: 0.39-0.95) <strong>and</strong> decreased CR-BSI (Level 1 outcome; RR<br />
0.56, 95% CI: 0.31-1). 61 However, the benefit <strong>of</strong> tunneling came primarily from one trial using<br />
the internal jugular as the site <strong>of</strong> catheter placement; the reduction in CR-BSI no longer reached<br />
statistical significance when data from the several subclavian catheter trials were pooled (RR<br />
0.71; 95% CI 0.36-1.43). 61 The authors concluded appropriately that current evidence does not
support the routine use <strong>of</strong> tunneling central venous catheters. This could change if the efficacy <strong>of</strong><br />
tunneling is clearly demonstrated at different placement sites <strong>and</strong> relative to other interventions<br />
(eg, antiseptic coated catheters). 61<br />
Ineffective <strong>Practices</strong><br />
Intravenous antimicrobial prophylaxis. There is no evidence to support the systemic use <strong>of</strong><br />
either vancomycin 62 or teicoplanin 63 during insertion <strong>of</strong> central venous catheters. The<br />
r<strong>and</strong>omized studies evaluating the use on intravenous vanomycin or teicoplanin have failed to<br />
demonstrate that this intervention reduces CR-BSI (Level 1 outcome). 62, 63 Given the theoretical<br />
risk <strong>of</strong> developing resistance to the antimicrobial agents used for prophylaxis, this practice is not<br />
recommended.<br />
References<br />
1. Raad II, Bodey GP. Infectious complications <strong>of</strong> indwelling vascular catheters. Clin Infect<br />
Dis 1992;15:197-208.<br />
2. Weinstein MP, Towns ML, Quartey SM, et al. The clinical significance <strong>of</strong> positive blood<br />
cultures in the 1990s: a prospective comprehensive evaluation <strong>of</strong> the microbiology,<br />
epidemiology, <strong>and</strong> outcome <strong>of</strong> bacteremia <strong>and</strong> fungemia in adults. Clin Infect Dis<br />
1997;24:584-602.<br />
3. Maki DG. Infections caused by intravascular devices used for infusion therapy. In: Bistro<br />
AI, F.A. W, eds. Infections Associated with Indwelling <strong>Medical</strong> Devices. Washington, DC:<br />
ASM Press, 1994:155-205.<br />
4. Mermel LA. Prevention <strong>of</strong> intravascular catheter-related infections. Ann Intern Med 2000;<br />
132:391-402.<br />
5. Guideline for prevention <strong>of</strong> intravascular device-related infections. Am J Infect Control<br />
1996; 24:261-293.<br />
6. Maki DG, Weise CE, Sarafin HW. A semiquantitative culture method for identifying<br />
intravenous-catheter-related infection. N Engl J Med 1977;296:1305-1309.<br />
7. Sherertz RJ, Heard SO, Raad, II. Diagnosis <strong>of</strong> triple-lumen catheter infection: comparison<br />
<strong>of</strong> roll plate, sonication, <strong>and</strong> flushing methodologies. J Clin Microbiol 1997;35:641-646.<br />
8. Flynn PM, Shenep JL, Barrett FF. Differential quantitation with a commercial blood<br />
culture tube for diagnosis <strong>of</strong> catheter-related infection. J Clin Microbiol 1988;26:1045-<br />
1046.<br />
9. Blot F, Nitenberg G, Chachaty E, et al. Diagnosis <strong>of</strong> catheter-related bacteraemia: a<br />
prospective comparison <strong>of</strong> the time to positivity <strong>of</strong> hub-blood versus peripheral-blood<br />
cultures. Lancet 1999;354:1071-1077.<br />
10. Bozzetti F, Terno G, Camerini E, Baticci F, Scarpa D, Pupa A. Pathogenesis <strong>and</strong><br />
predictability <strong>of</strong> central venous catheter sepsis. Surgery 1982;91:383-389.<br />
11. Hampton AA, Sherertz RJ. Vascular-access infections in hospitalized patients. Surg Clin<br />
North Am 1988;68:57-71.<br />
12. Corona ML, Peters SG, Narr BJ, Thompson RL. Infections related to central venous<br />
catheters . Mayo Clin Proc 1990;65:979-986.<br />
13. Snydman DR, Gorbea HF, Pober BR, Majka JA, Murray SA, Perry LK. Predictive value <strong>of</strong><br />
surveillance skin cultures in total-parenteral-nutrition-related infection. Lancet<br />
1982;2:1385-1388.<br />
179
14. Sitges Serra A, Puig P, Linares J, et al. Hub colonization as the initial step in an outbreak<br />
<strong>of</strong> catheter-related sepsis due to coagulase negative staphylococci during parenteral<br />
nutrition. JPEN J Parenter Enteral Nutr 1984;8:668-672.<br />
15. Kovacevich DS, Faubion WC, Bender JM, Schaberg DR, Wesley JR. Association <strong>of</strong><br />
parenteral nutrition catheter sepsis with urinary tract infections. JPEN J Parenter Enteral<br />
Nutr 1986;10:639-641.<br />
16. Maki DG, Rhame FS, Mackel DC, Bennett JV. Nationwide epidemic <strong>of</strong> septicemia caused<br />
by contaminated intravenous products. I. Epidemiologic <strong>and</strong> clinical features. Am J Med<br />
1976;60:471-485.<br />
17. Saint S, Veentra DL, Lipsky BA. The clinical <strong>and</strong> economic consequences <strong>of</strong> nosocomial<br />
central venous catheter-related infection: are antimicrobial catheters useful? Infect Control<br />
Hosp Epidemiol 2000;21:375-380.<br />
18. Anonymous. National Nosocomial Infections Surveillance (NNIS) System report, data<br />
summary from October 1986-April 1998, issued June 1998. Am J Infect Control<br />
1998;26:522-533.<br />
19. Pittet D, Tarara D, Wenzel RP. Nosocomial bloodstream infection in critically ill patients.<br />
Excess length <strong>of</strong> stay, extra costs, <strong>and</strong> attributable mortality. JAMA 1994;271:1598-601.<br />
20. Pittet D, Wenzel RP. Nosocomial bloodstream infections in the critically ill [letter]. JAMA<br />
1994;272:1819-1820.<br />
21. Smith RL, Meixler SM, Simberk<strong>of</strong>f MS. Excess mortality in critically ill patients with<br />
nosocomial bloodstream infections. Chest 1991;100:164-167.<br />
22. Digiovine B, Chenoweth C, Watts C, Higgins M. The attributable mortality <strong>and</strong> costs <strong>of</strong><br />
primary nosocomial bloodstream infections in the intensive care unit. Am J Respir Crit<br />
Care Med 1999;160:976-981.<br />
23. Soufir L, Timsit JF, Mahe C, Carlet J, Regnier B, Chevret S. Attributable morbidity <strong>and</strong><br />
mortality <strong>of</strong> catheter-related septicemia in critically ill patients: a matched, risk-adjusted,<br />
cohort study. Infect Control Hosp Epidemiol 1999;20:396-401.<br />
24. Cook D, R<strong>and</strong>olph A, Kernerman P, et al. Central venous catheter replacement strategies: a<br />
systematic review <strong>of</strong> the literature. Crit Care Med 1997;25:1417-1424.<br />
25. Eyer S, Brummitt C, Crossley K, Siegel R, Cerra F. Catheter-related sepsis: prospective,<br />
r<strong>and</strong>omized study <strong>of</strong> three methods <strong>of</strong> long-term catheter maintenance. Crit Care Med<br />
1990;18:1073-1079.<br />
26. Cobb DK, High KP, Sawyer RG, et al. A controlled trial <strong>of</strong> scheduled replacement <strong>of</strong><br />
central venous <strong>and</strong> pulmonary-artery catheters . N Engl J Med 1992;327:1062-8.<br />
27. Nystrom B. Impact <strong>of</strong> h<strong>and</strong>washing on mortality in intensive care: examination <strong>of</strong> the<br />
evidence. Infect Control Hosp Epidemiol 1994;15:435-436.<br />
28. Doebbeling BN, Stanley GL, Sheetz CT, et al. Comparative efficacy <strong>of</strong> alternative h<strong>and</strong>washing<br />
agents in reducing nosocomial infections in intensive care units. N Engl J Med<br />
1992;327:88-93.<br />
29. Raad, II, Hohn DC, Gilbreath BJ, et al. Prevention <strong>of</strong> central venous catheter-related<br />
infections by using maximal sterile barrier precautions during insertion. Infect Control<br />
Hosp Epidemiol 1994;15:231-238.<br />
30. Sherertz RJ, Ely EW, Westbrook DM, et al. Education <strong>of</strong> physicians-in-training can<br />
decrease the risk for vascular catheter infection. Ann Intern Med 2000;132:641-648.<br />
31. Tennenberg S, Lieser M, McCurdy B, et al. A prospective r<strong>and</strong>omized trial <strong>of</strong> an antibiotic<strong>and</strong><br />
antiseptic-coated central venous catheter in the prevention <strong>of</strong> catheter-related<br />
infections. Arch Surg 1997;132:1348-1351.<br />
180
32. Maki DG, Stolz SM, Wheeler S, Mermel LA. Prevention <strong>of</strong> central venous catheter-related<br />
bloodstream infection by use <strong>of</strong> an antiseptic-impregnated catheter. A r<strong>and</strong>omized,<br />
controlled trial. Ann Intern Med 1997;127:257-266.<br />
33. van Heerden PV, Webb SA, Fong S, Golledge CL, Roberts BL, Thompson WR. Central<br />
venous catheters revisited–infection rates <strong>and</strong> an assessment <strong>of</strong> the new Fibrin Analysing<br />
System brush. Anaesth Intensive Care 1996;24:330-333.<br />
34. Hannan M, Juste R, Shankar U, Nightingale C, Azadian B, Soni N. Colonization <strong>of</strong> triple<br />
lumen catheters. A study on antiseptic-bonded <strong>and</strong> st<strong>and</strong>ard catheters [abstract]. Clin<br />
Intensive Care 1996;7:56.<br />
35. Bach A, Bohrer H, Bottiger BW, Motsch J, Martin E. Reduction <strong>of</strong> bacterial colonization<br />
<strong>of</strong> triple-lumen catheters with antiseptic bonding in septic patients [abstract].<br />
Anesthesiology 1994;81:A261.<br />
36. Bach A, Schmidt H, Bottiger B, et al. Retention <strong>of</strong> antibacterial activity <strong>and</strong> bacterial<br />
colonization <strong>of</strong> antiseptic-bonded central venous catheters. J Antimicrob Chemother<br />
1996;37:315-322.<br />
37. Heard SO, Wagle M, Vijayakumar E, et al. Influence <strong>of</strong> triple-lumen central venous<br />
catheters coated with chlorhexidine <strong>and</strong> silver sulfadiazine on the incidence <strong>of</strong> catheterrelated<br />
bacteremia. Arch Intern Med 1998;158:81-87.<br />
38. Collin GR. Decreasing catheter colonization through the use <strong>of</strong> an antiseptic-impregnated<br />
catheter: a continuous quality improvement project. Chest 1999;115:1632-1640.<br />
39. Ciresi DL, Albrecht RM, Volkers PA, Scholten DJ. Failure <strong>of</strong> antiseptic bonding to prevent<br />
central venous catheter-related infection <strong>and</strong> sepsis. Am Surg 1996;62:641-646.<br />
40. Pemberton LB, Ross V, Cuddy P, Kremer H, Fessler T, McGurk E. No difference in<br />
catheter sepsis between st<strong>and</strong>ard <strong>and</strong> antiseptic central venous catheters. A prospective<br />
r<strong>and</strong>omized trial. Arch Surg 1996;131:986-989.<br />
41. Ramsay J, Nolte F, Schwarmann S. Incidence <strong>of</strong> catheter colonization <strong>and</strong> catheter-related<br />
infection with an antiseptic-impregnated triple lumen catheter [abstract]. Crit Care Med<br />
1994;22:A115.<br />
42. Trazzera S, Stern G, Rakesh B, Sinha S, Reiser P. Examination <strong>of</strong> antimicrobial-coated<br />
central venous catheters in patients at high risk for catheter-related infections in a medical<br />
intensive care unit <strong>and</strong> leukemia/bone marrow transplant unit [abstract]. Crit Care Med<br />
1995;23:A152.<br />
43. George SJ, Vuddamalay P, Boscoe MJ. Antiseptic-impregnated central venous catheters<br />
reduce the incidence <strong>of</strong> bacterial colonization <strong>and</strong> associated infection in<br />
immunocompromised transplant patients. Eur J Anaesthesiol 1997;14:428-431.<br />
44. Veenstra DL, Saint S, Saha S, Lumley T, Sullivan SD. Efficacy <strong>of</strong> antiseptic-impregnated<br />
central venous catheters in preventing catheter-related bloodstream infection: a metaanalysis.<br />
JAMA 1999;281:261-267.<br />
45. Raad I, Darouiche R, Dupuis J, et al. Central venous catheters coated with minocycline <strong>and</strong><br />
rifampin for the prevention <strong>of</strong> catheter-related colonization <strong>and</strong> bloodstream infections. A<br />
r<strong>and</strong>omized, double-blind trial. The Texas <strong>Medical</strong> Center Catheter Study Group. Ann<br />
Intern Med 1997;127:267-274.<br />
46. Darouiche RO, Raad, II, Heard SO, et al. A comparison <strong>of</strong> two antimicrobial-impregnated<br />
central venous catheters. Catheter Study Group. N Engl J Med 1999;340:1-8.<br />
47. Veenstra DL, Saint S, Sullivan SD. Cost-effectiveness <strong>of</strong> antiseptic-impregnated central<br />
venous catheters for the prevention <strong>of</strong> catheter-related bloodstream infection. JAMA<br />
1999;282:554-560.<br />
181
48. Maki DG, Ringer M, Alvarado CJ. Prospective r<strong>and</strong>omised trial <strong>of</strong> povidone-iodine,<br />
alcohol, <strong>and</strong> chlorhexidine for prevention <strong>of</strong> infection associated with central venous <strong>and</strong><br />
arterial catheters. Lancet 1991;338:339-343.<br />
49. Sheehan G, Leicht K, O'Brien M, Taylor G, Rennie R. Chlorhexidine versus povidineiodine<br />
as cutaneous antisepsis for prevention <strong>of</strong> vascular-catheter infection [abstract].<br />
Program <strong>and</strong> abstracts <strong>of</strong> the Interscience Conference on Antimicrobial Agents <strong>and</strong><br />
Chemotherapy 1993;33:A1616.<br />
50. Meffre C, Girard R, Hajjar J, Fabry J. Is peripheral venous catheter colonisation related to<br />
antiseptic used for disinfection <strong>of</strong> the insertion site? Hygiene 1995;9:45.<br />
51. Mimoz O, Pieroni L, Lawrence C, et al. Prospective, r<strong>and</strong>omized trial <strong>of</strong> two antiseptic<br />
solutions for prevention <strong>of</strong> central venous or arterial catheter colonization <strong>and</strong> infection in<br />
intensive care unit patients. Crit Care Med 1996;24:1818-1823.<br />
52. Cobett S, LeBlanc A. IV site infection: a prospective, r<strong>and</strong>omized clinical trial comparing<br />
the efficacy <strong>of</strong> three methods <strong>of</strong> skin antisepsis: CINA conference '99. CINA: Official<br />
Journal <strong>of</strong> the Canadian Intravenous Nurses Association 1999;15:48-49.<br />
53. Humar A, Ostromecki A, Direnfeld J, et al. Prospective r<strong>and</strong>omized trial <strong>of</strong> 10% povidoneiodine<br />
versus 0.5% tincture <strong>of</strong> chlorhexidine as cutaneous antisepsis for prevention <strong>of</strong><br />
central venous catheter infection. Clin Infect Dis 2000;31:1001-1007.<br />
54. Yusuf S. Obtaining medically meaningful answers from an overview <strong>of</strong> r<strong>and</strong>omized<br />
clinical trials. Stat Med 1987;6:281-294.<br />
55. Light RJ. Accumulating evidence from independent studies: what we can win <strong>and</strong> what we<br />
can lose. Stat Med 1987;6:221-231.<br />
56. Hennekens CH, Buring JE, Hebert PR. Implications <strong>of</strong> overviews <strong>of</strong> r<strong>and</strong>omized trials. Stat<br />
Med 1987;6:397-409.<br />
57. Sacks HS, Berrier J, Reitman D, Ancona Berk VA, Chalmers TC. Meta-analyses <strong>of</strong><br />
r<strong>and</strong>omized controlled trials. N Engl J Med 1987;316:450-455.<br />
58. Gelber RD, Goldhirsch A. Interpretation <strong>of</strong> results from subset analyses within overviews<br />
<strong>of</strong> r<strong>and</strong>omized clinical trials. Stat Med 1987;6:371-388.<br />
59. R<strong>and</strong>olph AG, Cook DJ, Gonzales CA, Andrew M. Benefit <strong>of</strong> heparin in central venous<br />
<strong>and</strong> pulmonary artery catheters: a meta-analysis <strong>of</strong> r<strong>and</strong>omized controlled trials. Chest<br />
1998;113:165-171.<br />
60. Mermel LA, McCormick RD, Springman SR, Maki DG. The pathogenesis <strong>and</strong><br />
epidemiology <strong>of</strong> catheter-related infection with pulmonary artery Swan-Ganz catheters: a<br />
prospective study utilizing molecular subtyping. Am J Med 1991;91:197S-205S.<br />
61. R<strong>and</strong>olph AG, Cook DJ, Gonzales CA, Brun-Buisson C. Tunneling short-term central<br />
venous catheters to prevent catheter-related infection: a meta-analysis <strong>of</strong> r<strong>and</strong>omized,<br />
controlled trials. Crit Care Med 1998;26:1452-1457.<br />
62. McKee R, Dunsmuir R, Whitby M, Garden OJ. Does antibiotic prophylaxis at the time <strong>of</strong><br />
catheter insertion reduce the incidence <strong>of</strong> catheter-related sepsis in intravenous nutrition? J<br />
Hosp Infect 1985;6:419-425.<br />
63. Ljungman P, Hagglund H, Bjorkstr<strong>and</strong> B, Lonnqvist B, Ringden O. Peroperative<br />
teicoplanin for prevention <strong>of</strong> gram-positive infections in neutropenic patients with<br />
indwelling central venous catheters: a r<strong>and</strong>omized, controlled study. Support Care Cancer<br />
1997;5:485-488.<br />
182
Final Comment to Chapter 16<br />
Infections due to central venous catheters are common <strong>and</strong> lead to substantial morbidity<br />
<strong>and</strong> health care costs. Several practices will likely reduce the incidence <strong>of</strong> this common patient<br />
safety problem, including the use <strong>of</strong> maximum sterile barrier precautions during catheter<br />
insertion, use <strong>of</strong> catheters coated with antibacterial or antiseptic agents, <strong>and</strong> use <strong>of</strong> chlorhexidine<br />
gluconate at the insertion site. Additionally, use <strong>of</strong> heparin <strong>and</strong> tunneling <strong>of</strong> the central venous<br />
catheter may prove to be effective in reducing CR-BSI. However, the relative efficacy <strong>of</strong> these<br />
interventions is unclear. Also, a clear <strong>and</strong> formal delineation <strong>of</strong> the economic consequences <strong>of</strong><br />
combining several <strong>of</strong> these patient safety practices is necessary.<br />
183
184
Chapter 17. Prevention <strong>of</strong> Ventilator-Associated Pneumonia<br />
Harold R Collard, MD<br />
University <strong>of</strong> Colorado Health Sciences Center<br />
Sanjay Saint, MD, MPH<br />
University <strong>of</strong> Michigan School <strong>of</strong> Medicine<br />
Introduction<br />
Ventilator-associated pneumonia (VAP) is a leading cause <strong>of</strong> morbidity <strong>and</strong> mortality in<br />
the intensive care unit (ICU). 1 The incidence <strong>of</strong> VAP varies greatly, ranging from 6 to 52% <strong>of</strong><br />
intubated patients depending on patient risk factors. The cumulative incidence is approximately<br />
1-3% per day <strong>of</strong> intubation. Overall, VAP is associated with an attributable mortality <strong>of</strong> up to<br />
30%. Attributable mortality approaches 50% when VAP is caused by the more virulent<br />
organisms that typify late-onset VAP (occurring 4 or more days into mechanical ventilation).<br />
The cost per episode <strong>of</strong> VAP is substantial, although specific data are lacking. The average cost<br />
per episode <strong>of</strong> nosocomial pneumonia is estimated at $3000 to $6000, <strong>and</strong> the additional length<br />
<strong>of</strong> stay for patients who develop VAP is estimated at 13 days. 1,2<br />
VAP is typically categorized as either early-onset VAP (occurring in the first 3-4 days <strong>of</strong><br />
mechanical ventilation) or late-onset VAP. This distinction is important microbiologically.<br />
Early-onset VAP is commonly caused by antibiotic-sensitive community-acquired organisms<br />
(eg, Streptococcus pneumoniae, Haemophilus influenzae, <strong>and</strong> Staphylococcus aureus). Lateonset<br />
VAP is commonly caused by antibiotic-resistant nosocomial organisms (eg, Pseudomonas<br />
aeruginosa, methicillin-resistant Staphylococcus aureus, Acinetobacter species, <strong>and</strong><br />
Enterobacter species). Most episodes <strong>of</strong> ventilator-associated pneumonia (VAP) are thought to<br />
develop from the aspiration <strong>of</strong> oropharyngeal secretions containing potentially pathogenic<br />
organisms. Aspiration <strong>of</strong> gastric secretions may also contribute, though likely to a lesser degree.<br />
Tracheal intubation interrupts the body’s anatomic <strong>and</strong> physiologic defenses against aspiration,<br />
making mechanical ventilation a major risk factor for VAP.<br />
This chapter reviews 4 practices that carry the potential to reduce the incidence <strong>of</strong> VAP<br />
in patients receiving mechanical ventilation. They are: variation in patient positioning,<br />
continuous aspiration <strong>of</strong> subglottic secretions, selective digestive tract decontamination, <strong>and</strong> the<br />
use <strong>of</strong> sucralfate.<br />
Subchapter 17.1. <strong>Patient</strong> Positioning: Semi-recumbent Positioning <strong>and</strong> Continuous<br />
Oscillation<br />
Background<br />
Aspiration <strong>of</strong> gastric secretions likely contributes to the development <strong>of</strong> VAP. 1 Semirecumbent<br />
positioning <strong>of</strong> mechanically ventilated patients may help reduce the incidence <strong>of</strong><br />
gastroesophogeal reflux <strong>and</strong> lead to a decreased incidence <strong>of</strong> VAP. Immobility in critically ill<br />
patients leads to atelectasis <strong>and</strong> decreased clearance <strong>of</strong> bronchopulmonary secretions. Both <strong>of</strong><br />
these sequelae may lead to increased risk <strong>of</strong> VAP. Continuous rotation <strong>and</strong> movement <strong>of</strong><br />
critically ill patients (termed continuous oscillation) may thus help prevent such changes.<br />
185
Semi-recumbent positioning<br />
Practice Description<br />
Semi-recumbent positioning is generally defined as elevation <strong>of</strong> the head <strong>of</strong> the bed to 45<br />
degrees. This is generally achieved in a hospital bed with patients’ feet remaining parallel to the<br />
floor (ie, the entire bed is not tilted) but this is not explicitly described in the published trials.<br />
Semi-recumbency is generally continued for the duration <strong>of</strong> mechanical ventilation.<br />
Opportunities for Impact<br />
Outside <strong>of</strong> select medical centers that have studied this practice, semi-recumbent<br />
positioning has not been widely adopted as the st<strong>and</strong>ard <strong>of</strong> care. Thus, such an intervention<br />
would have enormous opportunity for impact should it prove beneficial.<br />
Study Designs<br />
There have been three trials <strong>of</strong> semi-recumbent patient positioning <strong>and</strong> its effect on the<br />
incidence <strong>of</strong> VAP. 3-5 Two <strong>of</strong> these studies measured aspiration events using nuclear medicine<br />
techniques, the other was a r<strong>and</strong>omized trial with the primary outcome being VAP. In the one<br />
r<strong>and</strong>omized trial, 86 patients were r<strong>and</strong>omized at the time <strong>of</strong> intubation to semi-recumbent body<br />
position (45 degrees) or supine body position (0 degrees). 3 All patients received the same general<br />
critical care (eg, sterile endotracheal suctioning, stress ulcer prophylaxis with sucralfate if<br />
tolerating oral medications, no ventilator tubing changes, no selective digestive tract<br />
decontamination).<br />
Study Outcomes<br />
In the one r<strong>and</strong>omized clinical trial, VAP was clinically defined as a new <strong>and</strong> persistent<br />
infiltrate on chest radiography, plus two <strong>of</strong> the following: temperature <strong>of</strong> >38.3C, leukocyte<br />
count >12,000/mm 3 or
Potential for Harm<br />
No adverse effects were observed in patients r<strong>and</strong>omized to semi-recumbent positioning. 3<br />
However, patients were excluded if they had any <strong>of</strong> the following conditions: recent abdominal<br />
or neurologic surgery (
Potential for Harm<br />
There were no significant risks <strong>of</strong> continuous oscillation in any <strong>of</strong> the r<strong>and</strong>omized trials.<br />
Inadvertent disconnection <strong>of</strong> intravenous lines, increased ventricular ectopy, <strong>and</strong> patient<br />
intolerance were reported, but not quantified. Conscious patients tolerated the procedure poorly.<br />
Costs <strong>and</strong> Implementation<br />
The incremental cost <strong>of</strong> specialized beds capable <strong>of</strong> continuous oscillation has been<br />
estimated at approximately $100 per day. 9 A significant reduction in VAP incidence <strong>and</strong> length<br />
<strong>of</strong> stay could result in cost savings.<br />
Comment<br />
Both semi-recumbent positioning <strong>and</strong> continuous oscillation are relatively low-cost, lowrisk<br />
interventions. The one r<strong>and</strong>omized trial to date <strong>of</strong> semi-recumbent positioning shows it to be<br />
an effective method <strong>of</strong> reducing VAP. While it has not proven to provide a mortality benefit,<br />
semi-recumbent positioning is a safe <strong>and</strong> straightforward intervention whose effectiveness<br />
should be confirmed by additional r<strong>and</strong>omized clinical trials. Continuous oscillation is less<br />
clearly beneficial, although it may be effective in certain subgroups <strong>of</strong> patients (eg, surgical,<br />
neurologic). It also deserves continued study.<br />
188
Table 17.1.1. <strong>Patient</strong> positioning*<br />
Study Design Study<br />
Design,<br />
Semi-recumbent positioning<br />
R<strong>and</strong>omized controlled trial <strong>of</strong> semi-recumbent<br />
patient positioning in 86 mechanically ventilated<br />
patients. Primary outcome was VAP. (Drakulovic,<br />
1999) 3<br />
Two-period crossover trial <strong>of</strong> semi-recumbent<br />
patient positioning in 15 mechanically ventilated<br />
patients. Primary outcome was pulmonary<br />
aspiration. (Orozco-Levi, 1995) 4<br />
R<strong>and</strong>omized two-period crossover trial <strong>of</strong> semirecumbent<br />
patient positioning in 15 mechanically<br />
ventilated patients. Primary outcome was<br />
pulmonary aspiration. (Torres, 1992) 5<br />
Continuous oscillation<br />
R<strong>and</strong>omized controlled trial <strong>of</strong> continuous<br />
oscillation in 103 critically ill medical <strong>and</strong> surgical<br />
patients (90% mechanically ventilated). Primary<br />
outcomes included pneumonia. (Traver, 1995) 8<br />
Meta-analysis <strong>of</strong> 6 r<strong>and</strong>omized controlled trials <strong>of</strong><br />
continuous oscillation in critically ill surgical or<br />
stroke patients (majority mechanically ventilated).<br />
(Choi, 1992) 7<br />
R<strong>and</strong>omized controlled trial <strong>of</strong> continuous<br />
oscillation in 86 critically ill medical patients<br />
(majority mechanically ventilated). Primary<br />
outcomes included pneumonia. (Summer, 1989) 9<br />
189<br />
Outcomes<br />
Level 1,<br />
Level 1<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
Level 1,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
* RR indicates relative risk; VAP, ventilator-associated pneumonia.<br />
Pneumonia<br />
or<br />
Aspiration<br />
RR 0.24<br />
(p=0.003)<br />
RR 0.65<br />
(p0.05)<br />
No significant<br />
difference<br />
(data not<br />
reported)<br />
RR 0.93<br />
(p>0.05)
References<br />
1. Craven DE, Steger KA. Nosocomial pneumonia in mechanically ventilated adult patients:<br />
epidemiology <strong>and</strong> prevention in 1996. Semin Respir Infect. 1996;11:32-53.<br />
2. Thompson R. Prevention <strong>of</strong> nosocomial pneumonia. Med Clin North Am. 1994;78:1185-<br />
1198.<br />
3. Drakulovic MB, Torres A, Bauer TT, Nicolas JM, Nogue S, Ferrer M. Supine body<br />
position as a risk factor for nosocomial pneumonia in mechanically ventilated patients: a<br />
r<strong>and</strong>omised trial. Lancet. 1999;354:1851-1858.<br />
4. Orozco-Levi M, Torres A, Ferrer M, Piera C, el-Ebiary M, de la Bellacasa JP, et al.<br />
Semirecumbent position protects from pulmonary aspiration but not completely from<br />
gastroesophageal reflux in mechanically ventilated patients. Am J Respir Crit Care Med.<br />
1995;152:1387-1390.<br />
5. Torres A, Serra-Batlles J, Ros E, Piera C, Puig de la Bellacasa J, Cobos A, et al. Pulmonary<br />
aspiration <strong>of</strong> gastric contents in patients receiving mechanical ventilation: the effect <strong>of</strong><br />
body position. Ann Intern Med. 1992;116:540-543.<br />
6. Kollef MH. Ventilator-associated pneumonia. A multivariate analysis. JAMA.<br />
1993;270:1965-1970.<br />
7. Choi SC, Nelson LD. Kinetic Therapy in <strong>Critical</strong>ly Ill <strong>Patient</strong>s: Combined Results Based<br />
on Meta-<strong>Analysis</strong>. J Crit Care. 1992;7:57-62.<br />
8. Traver GA, Tyler ML, Hudson LD, Sherrill DL, Quan SF. Continuous oscillation: outcome<br />
in critically ill patients. J Crit Care. 1995;10:97-103.<br />
9. Summer WR, Curry P, Haponik EF, Nelson S, Elston R. Continuous Mechanical Turning<br />
<strong>of</strong> Intensive Care Unit <strong>Patient</strong>s Shortens Length <strong>of</strong> Stay in Some Diagnostic-Related<br />
Groups. J Crit Care. 1989;4:45-53.<br />
Subchapter 17.2. Continuous Aspiration <strong>of</strong> Subglottic Secretions<br />
Background<br />
Ventilator-associated pneumonia (VAP) frequently develops from the aspiration <strong>of</strong><br />
oropharyngeal secretions containing potentially pathogenic organisms. 1 Tracheal intubation<br />
interrupts the body’s anatomic <strong>and</strong> physiologic defenses against aspiration, making mechanical<br />
ventilation a major risk factor for VAP. The accumulation <strong>of</strong> contaminated oropharyngeal<br />
secretions above the endotracheal tube cuff may contribute to the risk <strong>of</strong> aspiration. 1 Removal <strong>of</strong><br />
these pooled secretions through suctioning <strong>of</strong> the subglottic region, termed continuous aspiration<br />
<strong>of</strong> subglottic secretions (CASS), may reduce the risk <strong>of</strong> developing VAP.<br />
Practice Description<br />
Continuous aspiration <strong>of</strong> subglottic secretions requires intubation with specially designed<br />
endotracheal tubes (see Figure 17.2.1). These endotracheal tubes contain a separate dorsal lumen<br />
that opens into the subglottic region, allowing for aspiration <strong>of</strong> any pooled secretions. The<br />
amount <strong>of</strong> secretions is monitored (usually daily) <strong>and</strong> the patency <strong>of</strong> the suction lumen is tested<br />
frequently (every few hours). In studies <strong>of</strong> the impact <strong>of</strong> this practice, aspiration has been applied<br />
from time <strong>of</strong> intubation to time <strong>of</strong> extubation. One <strong>of</strong> the studies tested manual aspiration<br />
performed hourly instead <strong>of</strong> continuous mechanical aspiration. 2<br />
190
Opportunities for Impact<br />
Continuous aspiration <strong>of</strong> subglottic secretions is an uncommon practice. The<br />
opportunities for impact are therefore significant should this practice prove beneficial in<br />
lowering rates <strong>of</strong> VAP.<br />
Study Designs<br />
There have been three r<strong>and</strong>omized controlled trials <strong>of</strong> CASS to date. 2-4 (Table 17.2.1)<br />
Two have included both medical <strong>and</strong> surgical patients requiring mechanical ventilation for<br />
greater than 72 hours <strong>and</strong> one included only post-cardiac surgery patients. All three studies<br />
r<strong>and</strong>omized patients to CASS or st<strong>and</strong>ard care. Attempts were made to control for additional,<br />
potentially effective preventive strategies such as patient positioning, frequency <strong>of</strong> ventilator<br />
circuit changes, type <strong>of</strong> stress ulcer prophylaxis used, <strong>and</strong> administration <strong>of</strong> antibiotics.<br />
Study Outcomes<br />
All trials reported development <strong>of</strong> VAP <strong>and</strong> mortality at the time <strong>of</strong> extubation, ICU or<br />
hospital discharge. VAP was generally defined as a new radiographic infiltrate plus two <strong>of</strong> the<br />
following: fever, leukocytosis/leukopenia, or purulent tracheal aspirate. Microbiologic<br />
confirmation was not consistently obtained. Time to development <strong>of</strong> VAP was also reported.<br />
Mortality was reported at time <strong>of</strong> discharge from the hospital.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
One <strong>of</strong> the three trials found a statistically significant decrease in the incidence <strong>of</strong> VAP<br />
with CASS when compared to st<strong>and</strong>ard treatment, while a second study showed a strong trend<br />
(See Table 17.2.1). 2,3 All three trials reported a statistically significant delay in the time to<br />
development <strong>of</strong> VAP, ranging from 48 hours to 8 days. Two trials found a decreased incidence<br />
<strong>of</strong> VAP caused by Staphylococcus aureus <strong>and</strong> Hemophilus influenzae, but no change was<br />
observed in the incidence <strong>of</strong> VAP caused by Pseudomonas aeruginosa or Enterobacteriaceae. 3,4<br />
No difference in mortality was observed in any <strong>of</strong> the trials.<br />
Potential for Harm<br />
There is minimal potential for harm to patients from the application <strong>of</strong> CASS <strong>and</strong> no<br />
adverse patient events were reported in over 150 patients. 4<br />
Costs <strong>and</strong> Implementation<br />
The cost <strong>and</strong> cost-effectiveness <strong>of</strong> CASS have not been examined. The direct costs<br />
appear minimal. Hi-Lo Evac tubes cost approximately 25% more than st<strong>and</strong>ard endotracheal<br />
tubes, putting the estimated cost <strong>of</strong> each unit at less than $1. 2 The cost-savings per episode <strong>of</strong><br />
VAP prevented could therefore be substantial. Implementation would largely be a matter <strong>of</strong><br />
making the specialized endotracheal tubes available <strong>and</strong> providing staff training. The mechanical<br />
suctioning apparatus would require frequent monitoring by nursing or respiratory therapy to<br />
insure adequate function.<br />
Comment<br />
Continuous aspiration <strong>of</strong> subglottic secretions is a promising strategy for the prevention<br />
<strong>of</strong> VAP. Two r<strong>and</strong>omized controlled trials have suggested a decrease in the rate <strong>of</strong> VAP in<br />
patients requiring prolonged (>3 days) mechanical ventilation (only one trial was statistically<br />
191
significant). The third trial showed no difference, but the patient population in this trial included<br />
many short-term intubations (mean duration <strong>of</strong> 36 hours) <strong>and</strong> was restricted to patients<br />
undergoing cardiac surgery. Larger r<strong>and</strong>omized controlled trials are needed to address the impact<br />
<strong>of</strong> CASS more definitively.<br />
Another interesting observation is the delay in the development <strong>of</strong> VAP <strong>and</strong> the<br />
decreased incidence <strong>of</strong> Staphylococcus aureus <strong>and</strong> Hemophilus influenzae. This suggests that<br />
CASS may provide most <strong>of</strong> its benefit by preventing early VAP caused by community-acquired<br />
organisms, <strong>and</strong> its use could therefore be targeted to those patients requiring mechanical<br />
ventilation for intermediate periods <strong>of</strong> time (ie, those at greatest risk for early VAP).<br />
Figure 17.2.1. Diagram <strong>of</strong> continuous aspiration <strong>of</strong> subglottic secretions (copied with<br />
permission) 3<br />
192
Table 17.2.1. R<strong>and</strong>omized trials <strong>of</strong> continuous aspiration <strong>of</strong> subglottic secretions*<br />
Study Description Study<br />
Outcomes<br />
Kollef, 1999 4 : 343 patients undergoing<br />
cardiac surgery <strong>and</strong> requiring mechanical<br />
ventilation<br />
Valles, 1995 3 : 153 patients requiring<br />
prolonged mechanical ventilation<br />
Mahul, 1992 2 : 145 patients requiring<br />
mechanical ventilation for more than 3 days<br />
* CI indicates confidence interval.<br />
193<br />
Relative Risk <strong>of</strong><br />
Pneumonia<br />
(95% CI)<br />
Relative Risk<br />
<strong>of</strong> Mortality<br />
(95% CI)<br />
Level 1 0.61 (0.27-1.40) 0.86 (0.30-2.42)<br />
Level 1 0.47 (0.21-1.06) 1.09 (0.72-1.63)<br />
Level 1 0.46 (0.23-0.93) 1.14 (0.62-2.07)<br />
References<br />
1. Craven DE, Steger KA. Nosocomial pneumonia in mechanically ventilated adult patients:<br />
epidemiology <strong>and</strong> prevention in 1996. Semin Respir Infect. 1996;11:32-53.<br />
2. Mahul P, Auboyer C, Jospe R, Ros A, Guerin C, el Khouri Z, et al. Prevention <strong>of</strong><br />
nosocomial pneumonia in intubated patients: respective role <strong>of</strong> mechanical subglottic<br />
secretions drainage <strong>and</strong> stress ulcer prophylaxis. Intensive Care Med. 1992;18:20-25.<br />
3. Valles J, Artigas A, Rello J, Bonsoms N, Fontanals D, Blanch L, et al. Continuous<br />
aspiration <strong>of</strong> subglottic secretions in preventing ventilator- associated pneumonia. Ann<br />
Intern Med. 1995;122:179-186.<br />
4. Kollef MH, Skubas NJ, Sundt TM. A r<strong>and</strong>omized clinical trial <strong>of</strong> continuous aspiration <strong>of</strong><br />
subglottic secretions in cardiac surgery patients. Chest. 1999;116:1339-1346.<br />
Subchapter 17.3. Selective Digestive Tract Decontamination<br />
Background<br />
Selective digestive tract decontamination (SDD) involves the use <strong>of</strong> non-absorbable<br />
antibiotics topically applied to the gastrointestinal tract in an effort to sterilize the oropharynx<br />
<strong>and</strong> stomach. The goal is to decrease the pathogenicity <strong>of</strong> aspirated secretions <strong>and</strong> thereby reduce<br />
the incidence <strong>of</strong> VAP.<br />
Practice Description<br />
Most studies have used a combination <strong>of</strong> topical polymixin, tobramycin or gentamicin,<br />
<strong>and</strong> amphotericin applied to the oropharynx (by h<strong>and</strong>) <strong>and</strong> the stomach (by nasogastric tube). 1<br />
About half <strong>of</strong> the studies also included a short (3-4 day) course <strong>of</strong> systemic intravenous<br />
antimicrobial therapy, most commonly ceftriaxone. In general, topical antibiotics were applied<br />
several times daily from the time <strong>of</strong> intubation until extubation (or shortly thereafter).
Opportunities for Impact<br />
SDD is not widely used in the United States. 1 The Centers for Disease Control <strong>and</strong><br />
Prevention <strong>and</strong> the American Thoracic Society's guidelines published in the 1990s do not<br />
recommend its routine use. 2,3 Given the frequency <strong>and</strong> morbidity <strong>of</strong> VAP, if the practice is<br />
beneficial substantial opportunity for patient safety enhancement exists.<br />
Study Designs<br />
There have been over 30 r<strong>and</strong>omized controlled trials <strong>and</strong> seven meta-analyses <strong>of</strong> SDD<br />
(see Table 17.3.1). 4-10 A representative meta-analysis identified 33 r<strong>and</strong>omized trials <strong>of</strong> SDD<br />
using a structured search <strong>of</strong> the literature that met the authors’ methodologic inclusion criteria:<br />
measurement <strong>of</strong> clinical outcomes (including VAP <strong>and</strong> mortality), inclusion <strong>of</strong> unselected patient<br />
populations, <strong>and</strong> mechanical ventilation in at least half <strong>of</strong> patients. 1 As with several <strong>of</strong> the other<br />
meta-analyses, individual trials in this particular meta-analysis were grouped into those that used<br />
topical antibiotics only <strong>and</strong> those that used topical <strong>and</strong> systemic antibiotics. This meta-analysis<br />
was unique in that the investigators obtained individual patient data for the majority <strong>of</strong> patients<br />
(4343 (76%) <strong>of</strong> the 5727 patients involved). 1<br />
Study Outcomes<br />
All meta-analyses reported risk <strong>of</strong> VAP <strong>and</strong> mortality at hospital or ICU discharge.<br />
Individual study outcomes also included number <strong>of</strong> days intubated, length <strong>of</strong> ICU stay, duration<br />
<strong>of</strong> antibiotic therapy, time to onset <strong>of</strong> VAP, <strong>and</strong> cost. Several meta-analyses performed subgroup<br />
analysis to assess the importance <strong>of</strong> statistical methods (eg, quality <strong>of</strong> r<strong>and</strong>omization, blinding,<br />
VAP definition) <strong>and</strong> clinical factors (eg, Acute Physiology <strong>and</strong> Chronic Health Evaluation<br />
(APACHE) score).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
All seven meta-analyses report substantial reduction in the risk <strong>of</strong> VAP with the use <strong>of</strong><br />
SDD (see Table 17.3.1). Four <strong>of</strong> seven meta-analyses report a statistically significant reduction<br />
in mortality. 4-6,8 Four <strong>of</strong> seven meta-analyses separately analyzed trials using topical antibiotics<br />
only <strong>and</strong> those using topical <strong>and</strong> systemic antibiotics. 4,6,8,9 All four revealed a statistically<br />
significant mortality benefit with combined topical <strong>and</strong> systemic prophylaxis <strong>and</strong> no mortality<br />
benefit with topical prophylaxis alone. 1,6,8,9 However, these four meta-analyses did reveal a<br />
significant decrease in VAP incidence in those given topical antibiotics only compared to the<br />
placebo group. 4,6,8,9 Several <strong>of</strong> the meta-analyses included subgroup analyses to assess the<br />
benefit <strong>of</strong> SDD in patients categorized by type <strong>of</strong> illness (surgical, medical) <strong>and</strong> severity <strong>of</strong><br />
illness (APACHE score), with conflicting results. 4,6<br />
Potential for Harm<br />
There were no significant adverse events reported in most trials, although allergic<br />
reactions to the antibiotic preparations have been uncommonly noted. The primary long-term<br />
concern with the widespread use <strong>of</strong> SDD is the development <strong>of</strong> antibiotic resistance. 11-13 The<br />
data are unclear regarding the impact <strong>of</strong> SDD on the emergence <strong>of</strong> resistant organisms, <strong>and</strong> no<br />
study has demonstrated an impact <strong>of</strong> increased bacterial resistance on morbidity or mortality.<br />
Costs <strong>and</strong> Implementation<br />
The cost <strong>of</strong> implementing SDD appears minimal in most trials, but there have been no in<br />
depth reviews <strong>of</strong> the subject. Several trials have found that patients receiving SDD had lower<br />
194
total antibiotic costs. 14-16 Overall hospital costs also may be lower, mediated through the<br />
decreased rate <strong>of</strong> VAP. 17<br />
Comment<br />
SDD is a very promising method <strong>of</strong> reducing VAP <strong>and</strong> ICU-related mortality. The data<br />
supporting a significant reduction in risk <strong>of</strong> VAP <strong>and</strong> short-term mortality with SDD using<br />
topical <strong>and</strong> short-term intravenous antibiotics are strong. SDD is a relatively non-invasive<br />
intervention <strong>and</strong> the additional financial cost is minimal. What remains to be determined is the<br />
long-term effect <strong>of</strong> SDD on antibiotic resistance patterns, <strong>and</strong> the impact <strong>of</strong> such effect on<br />
morbidity <strong>and</strong> mortality. Research into the impact <strong>of</strong> SDD on the emergence <strong>of</strong> antibiotic<br />
resistance should be strongly encouraged.<br />
195
Table 17.3.1. Meta-analyses <strong>of</strong> selective digestive tract decontamination*<br />
Study Design Pneumonia (95% CI) Mortality (95% CI)<br />
Nathens, 1999 6 : 21 r<strong>and</strong>omized<br />
controlled trials <strong>of</strong> antibiotic<br />
prophylaxis used to decrease<br />
nosocomial respiratory tract infections;<br />
dual analysis <strong>of</strong> medical <strong>and</strong> surgical<br />
patients<br />
D’Amico, 1998 1 : 33 r<strong>and</strong>omized<br />
controlled trials from <strong>of</strong> antibiotic<br />
prophylaxis used to decrease<br />
nosocomial respiratory tract infections;<br />
dual analysis <strong>of</strong> topical <strong>and</strong> systemic<br />
antibiotics combined <strong>and</strong> topical<br />
antibiotics alone<br />
Hurley, 1995 5 : 26 r<strong>and</strong>omized<br />
controlled trials <strong>of</strong> antibiotic<br />
prophylaxis used to decrease<br />
nosocomial respiratory tract infections<br />
Kollef, 1994 7 : 16 r<strong>and</strong>omized<br />
controlled trials <strong>of</strong> antibiotic<br />
prophylaxis used to decrease<br />
nosocomial respiratory tract infections<br />
Heyl<strong>and</strong>, 1994 8 : 25 r<strong>and</strong>omized<br />
controlled trials <strong>of</strong> antibiotic<br />
prophylaxis used to decrease<br />
nosocomial respiratory tract infections;<br />
performed subgroup analyses<br />
SDD Trialists’ Collaborative Group,<br />
1993 9 : 22 r<strong>and</strong>omized controlled trials<br />
<strong>of</strong> antibiotic prophylaxis used to<br />
decrease nosocomial respiratory tract<br />
infections; performed subgroup<br />
analyses<br />
V<strong>and</strong>enbroucke-Grauls, 1991 10 : 6<br />
r<strong>and</strong>omized controlled trials <strong>of</strong><br />
antibiotic prophylaxis used to decrease<br />
nosocomial respiratory tract infections<br />
<strong>Medical</strong>: OR 0.45 (0.33-<br />
0.62)<br />
Surgical: OR 0.19 (0.15-<br />
0.26)<br />
Overall: not reported<br />
Topical/IV: OR 0.35<br />
(0.29-0.41)<br />
Topical: OR 0.56 (0.46-<br />
0.68)<br />
Overall: OR 0.35 (0.30-<br />
0.42)<br />
Overall: RD 0.145<br />
(0.116-0.174)<br />
Overall: RR 0.46 (0.39-<br />
0.56)<br />
Topical/IV: RR 0.48<br />
(0.39-0.60)<br />
Topical: RR 0.43 (0.32-<br />
0.59)<br />
Overall: OR 0.37 (0.31-<br />
0.43)<br />
Topical/IV: OR 0.33<br />
(0.27-0.40)<br />
Topical: OR 0.43 (0.33-<br />
0.56)<br />
Overall: OR 0.12 (0.08-<br />
0.19)<br />
196<br />
<strong>Medical</strong><br />
Overall: OR 0.91 (0.71-1.18)<br />
Topical/IV: OR 0.75 (0.53-1.06)<br />
Topical: OR 1.14 (0.77-1.68)<br />
Surgical<br />
Overall: OR 0.70 (0.52-0.93)<br />
Topical/IV: OR 0.60 (0.41-0.88)<br />
Topical: OR 0.86 (0.51-1.45)<br />
Overall: OR 0.88 (0.78-0.98)<br />
Topical/IV: OR 0.80 (0.69-0.93)<br />
Topical: OR 1.01 (0.84-1.22)<br />
Overall: OR 0.86 (0.74-0.99)<br />
Overall: RD 0.019 (-0.016-0.054)<br />
Overall: RR 0.87 (0.79-0.97)<br />
Topical/IV: RR 0.81 (0.71-0.95)<br />
Topical: RR 1.00 (0.83-1.19)<br />
Overall: OR 0.90 (0.79-1.04)<br />
Topical/IV: OR 0.80 (0.67-0.97)<br />
Topical: OR 1.07 (0.86-1.32)<br />
Overall: OR 0.70 (0.45-1.09)<br />
* CI indicates confidence interval; RD, risk difference, RR, relative risk; <strong>and</strong> OR, odds ratio.
References<br />
1. D'Amico R, Pifferi S, Leonetti C, Torri V, Tinazzi A, Liberati A. Effectiveness <strong>of</strong><br />
antibiotic prophylaxis in critically ill adult patients: systematic review <strong>of</strong> r<strong>and</strong>omised<br />
controlled trials. BMJ. 1998;316:1275-1285.<br />
2. Hospital-acquired pneumonia in adults: diagnosis, assessment <strong>of</strong> severity, initial<br />
antimicrobial therapy, <strong>and</strong> preventive strategies. A consensus statement, American<br />
Thoracic Society, November 1995. Am J Respir Crit Care Med. 1996;153:1711-1725.<br />
3. CDC. Guidelines for prevention <strong>of</strong> nosocomial pneumonia. Centers for Disease Control<br />
<strong>and</strong> Prevention. MMWR Morb Mortal Wkly Rep. 1997;46:1-79.<br />
4. Craven DE, Steger KA. Nosocomial pneumonia in mechanically ventilated adult patients:<br />
epidemiology <strong>and</strong> prevention in 1996. Semin Respir Infect. 1996;11:32-53.<br />
5. Hurley JC. Prophylaxis with enteral antibiotics in ventilated patients: selective<br />
decontamination or selective cross-infection? Antimicrob Agents Chemother. 1995;39:941-<br />
947.<br />
6. Nathens AB, Marshall JC. Selective decontamination <strong>of</strong> the digestive tract in surgical<br />
patients: a systematic review <strong>of</strong> the evidence. Arch Surg. 1999;134:170-176.<br />
7. Kollef MH. The role <strong>of</strong> selective digestive tract decontamination on mortality <strong>and</strong><br />
respiratory tract infections. A meta-analysis. Chest. 1994;105:1101-1108.<br />
8. Heyl<strong>and</strong> DK, Cook DJ, Jaeschke R, Griffith L, Lee HN, Guyatt GH. Selective<br />
decontamination <strong>of</strong> the digestive tract. An overview. Chest. 1994;105:1221-1229.<br />
9. Meta-analysis <strong>of</strong> r<strong>and</strong>omised controlled trials <strong>of</strong> selective decontamination <strong>of</strong> the digestive<br />
tract. Selective Decontamination <strong>of</strong> the Digestive Tract Trialists' Collaborative Group.<br />
BMJ. 1993;307:525-532.<br />
10. V<strong>and</strong>enbroucke-Grauls CM, V<strong>and</strong>enbroucke JP. Effect <strong>of</strong> selective decontamination <strong>of</strong> the<br />
digestive tract on respiratory tract infections <strong>and</strong> mortality in the intensive care unit.<br />
Lancet. 1991;338:859-862.<br />
11. van Saene HK, Stoutenbeek CP, Hart CA. Selective decontamination <strong>of</strong> the digestive tract<br />
(SDD) in intensive care patients: a critical evaluation <strong>of</strong> the clinical, bacteriological <strong>and</strong><br />
epidemiological benefits. J Hosp Infect. 1991;18:261-277.<br />
12. Ebner W, Kropec-Hubner A, Daschner FD. Bacterial resistance <strong>and</strong> overgrowth due to<br />
selective decontamination <strong>of</strong> the digestive tract. Eur J Clin Microbiol Infect Dis.<br />
2000;19:243-247.<br />
13. Bartlett JG. Selective decontamination <strong>of</strong> the digestive tract <strong>and</strong> its effect on antimicrobial<br />
resistance. Crit Care Med. 1995;23:613-615.<br />
14. Quinio B, Albanese J, Bues-Charbit M, Vivi<strong>and</strong> X, Martin C. Selective decontamination <strong>of</strong><br />
the digestive tract in multiple trauma patients. A prospective double-blind, r<strong>and</strong>omized,<br />
placebo-controlled study. Chest. 1996;109:765-772.<br />
15. Verwaest C, Verhaegen J, Ferdin<strong>and</strong>e P, Schetz M, Van den Berghe G, Verbist L, et al.<br />
R<strong>and</strong>omized, controlled trial <strong>of</strong> selective digestive decontamination in 600 mechanically<br />
ventilated patients in a multidisciplinary intensive care unit. Crit Care Med. 1997;25:63-<br />
71.<br />
16. Sanchez Garcia M, Cambronero Galache JA, Lopez Diaz J, Cerda Cerda E, Rubio Blasco<br />
J, Gomez Aguinaga MA, et al. Effectiveness <strong>and</strong> cost <strong>of</strong> selective decontamination <strong>of</strong> the<br />
197
digestive tract in critically ill intubated patients. A r<strong>and</strong>omized, double-blind, placebocontrolled,<br />
multicenter trial. Am J Respir Crit Care Med. 1998;158:908-916.<br />
17. Silvestri L, Mannucci F, van Saene HK. Selective decontamination <strong>of</strong> the digestive tract: a<br />
life saver. J Hosp Infect. 2000;45:185-190.<br />
Subchapter 17.4. Sucralfate <strong>and</strong> Prevention <strong>of</strong> VAP<br />
Background<br />
Aspiration <strong>of</strong> gastric secretions may contribute to the development <strong>of</strong> VAP. It has been<br />
observed that gastric colonization by potentially pathogenic organisms increases with decreasing<br />
gastric acidity, leading to the hypothesis that pH-altering drugs may cause increased rates <strong>of</strong><br />
VAP. 1 H2-antagonist therapy, widely used in mechanically-ventilated patients for stress ulcer<br />
prophylaxis (see Chapter 34), significantly elevates gastric pH. Sucralfate, an alternative<br />
prophylactic agent that does not affect gastric pH, may allow less gastric colonization with<br />
potentially pathogenic organisms than H2-antagonists <strong>and</strong> therefore prevent some cases <strong>of</strong> VAP.<br />
Practice Description<br />
In general, 1 g <strong>of</strong> sucralfate suspension is given through a nasogastric tube every four to<br />
six hours. When H2-antagonists are used, their dosing <strong>and</strong> frequency vary. A representative study<br />
used 50 mg <strong>of</strong> ranitidine intravenously every eight hours, dose adjusted for creatinine clearance. 2<br />
Stress ulcer prophylaxis is usually started upon initiation <strong>of</strong> mechanical ventilation <strong>and</strong><br />
continued until extubation (Chapter 34).<br />
Opportunities for Impact<br />
Stress ulcer prophylaxis is usually given to critically ill ventilated patients. A large cohort<br />
study <strong>of</strong> over 2000 critically ill patients suggests that the majority receive H2-antagonists<br />
(71.8%) followed by sucralfate (7.0%) <strong>and</strong> combination therapy (15.4%) or other single agents<br />
(omeprazole, antacids, prostagl<strong>and</strong>ins). 3 There is, therefore, significant opportunity for impact<br />
should sucralfate prove to lower rates <strong>of</strong> VAP <strong>and</strong> improve survival.<br />
Study Designs<br />
There have been over 20 r<strong>and</strong>omized controlled trials <strong>of</strong> stress ulcer prophylaxis using<br />
sucralfate, H2-antagonists, <strong>and</strong> other therapies in critically ill patients. Seven meta-analyses have<br />
been published to date. 2,4-10 The individual trials <strong>and</strong> meta-analyses have significant variation in<br />
methodology. In general, the individual trials r<strong>and</strong>omized critically ill patients to sucralfate, H2antagonists,<br />
other agents such as antacids <strong>and</strong> pirenzepine, or placebo. The majority <strong>of</strong> patients<br />
included in these studies required mechanical ventilation. Various drug-drug, drug-placebo<br />
combinations were compared, <strong>and</strong> the rates <strong>of</strong> VAP <strong>and</strong> mortality were recorded.<br />
Study Outcomes<br />
Most trials report development <strong>of</strong> VAP <strong>and</strong> mortality as primary endpoints. There is<br />
significant variation in the definition <strong>of</strong> VAP used in these trials. In the largest <strong>and</strong> most recent<br />
r<strong>and</strong>omized trial, VAP was defined as a new radiographic infiltrate plus two <strong>of</strong> the following:<br />
temperature <strong>of</strong> >38.5°C or 10,000/mm 3 or
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The results <strong>of</strong> the seven meta-analyses <strong>and</strong> one recent large r<strong>and</strong>omized controlled trial<br />
are inconclusive (see Table 17.4.1). The two largest meta-analyses to date suggest a decreased<br />
incidence <strong>of</strong> VAP with sucralfate compared to H2-antagonists, 4,6 <strong>and</strong> one reports a statistically<br />
significant mortality benefit with sucralfate. 6 A recent r<strong>and</strong>omized controlled trial <strong>of</strong> 1200<br />
ventilated patients reports no significant difference between the two therapies in terms <strong>of</strong> VAP or<br />
mortality. 8<br />
Potential for Harm<br />
Sucralfate therapy has been associated with a statistically significant increased risk <strong>of</strong><br />
clinically important gastrointestinal bleeding when compared to H2-antagonists. 8 Clinically<br />
important bleeding developed in 3.8% <strong>of</strong> patients receiving sucralfate compared with 1.7% <strong>of</strong><br />
patients receiving H2-antagonists (relative risk for H2-antagonist = 0.44, 95% CI: 0.21-0.92). 8<br />
Gastrointestinal bleeding in critically ill patients has an attributable mortality <strong>of</strong> approximately<br />
12.5%. 8 While previous meta-analyses have suggested little difference in rates <strong>of</strong> gastrointestinal<br />
bleeding between the various prophylactic agents, 6 these results from a large r<strong>and</strong>omized trial are<br />
convincing. There are very few adverse effects from sucralfate therapy aside from constipation,<br />
rare nausea <strong>and</strong> vomiting <strong>and</strong> very rare bezoar formation <strong>and</strong> aluminum intoxication. 11 Sucralfate<br />
administration has been associated with transmission <strong>of</strong> vancomycin resistant enterococcus,<br />
likely due to increased manipulation <strong>of</strong> patients’ nasogastric tubes. 12 Unlike parenteral H2blockers,<br />
sucralfate m<strong>and</strong>ates nasogastric tube placement in intubated patients. The drug can also<br />
lead to decreased absorption <strong>of</strong> other medications.<br />
Costs <strong>and</strong> Implementation<br />
Several studies have looked at the cost-effectiveness <strong>of</strong> stress ulcer prophylaxis. 13,14<br />
Based on decision analysis, the cost per episode <strong>of</strong> gastrointestinal bleeding averted in high-risk<br />
patients is several thous<strong>and</strong> dollars greater with H2-antagonists than with sucralfate. 13 This cost<br />
difference remains significant even if H2-antagonists are assumed to be 50% more effective.<br />
There are no reliable data comparing overall costs from the actual clinical trials. The mean cost,<br />
largely driven by prolonged length <strong>of</strong> stay, is significantly higher for patients who bleed than for<br />
those who do not ($70,000 vs. $15-20,000), implying that in patients at high risk for GI bleeding<br />
(eg, mechanically ventilated patients, those with a coagulopathy), stress ulcer prophylaxis may<br />
be cost-neutral or even cost-saving (see also Chapter 34). 14 Implementation <strong>of</strong> sucralfate use<br />
would be largely an issue <strong>of</strong> staff education as administration is relatively uncomplicated.<br />
Comment<br />
The data supporting stress ulcer prophylaxis with sucralfate instead <strong>of</strong> H2-antagonists to<br />
prevent VAP are inconclusive, <strong>and</strong> the theoretical contribution <strong>of</strong> increased gastric colonization<br />
with potentially pathogenic organisms to the development <strong>of</strong> VAP is unproven. There are data<br />
both supporting <strong>and</strong> refuting a decreased incidence <strong>of</strong> VAP with sucralfate. Most investigators<br />
have found at least a trend toward decreased incidence <strong>of</strong> VAP with sucralfate, <strong>and</strong> larger studies<br />
are warranted. The greatest benefit from sucralfate may be the prevention <strong>of</strong> late-onset VAP in<br />
patients requiring long-term ventilation. 15 Any increased risk <strong>of</strong> gastrointestinal bleeding with<br />
sucralfate therapy in these patients may be <strong>of</strong>fset by the decreased risk <strong>of</strong> VAP. Until the data are<br />
more definitive, however, when stress ulcer prophylaxis is deemed appropriate (Chapter 34), the<br />
199
use <strong>of</strong> H2-blockers seems preferable to sucralfate because <strong>of</strong> the former’s superiority in<br />
preventing clinically important gastrointestinal bleeding.<br />
200
Table 17.4.1. Studies <strong>of</strong> stress ulcer prophylaxis<br />
Study Design Design,<br />
Outcomes<br />
Meta-analysis <strong>of</strong> r<strong>and</strong>omized<br />
controlled trials comparing<br />
ranitidine with placebo,<br />
sucralfate with placebo <strong>and</strong><br />
ranitidine with sucralfate for the<br />
prevention <strong>of</strong> pneumonia in<br />
critically ill patients. (Messori,<br />
2000)<br />
Multicenter r<strong>and</strong>omized, blinded,<br />
placebo-controlled trial <strong>of</strong><br />
sucralfate with ranitidine in 1200<br />
critically ill mechanically<br />
ventilated patients. Endpoints<br />
were gastrointestinal bleeding,<br />
VAP <strong>and</strong> mortality. (Cook,<br />
1998)<br />
27 r<strong>and</strong>omized trials <strong>of</strong> stress<br />
ulcer prophylaxis in critically ill<br />
patients. The majority <strong>of</strong> patients<br />
were mechanically ventilated.<br />
Endpoints were gastrointestinal<br />
bleeding, pneumonia <strong>and</strong><br />
mortality. (Cook, 1996)<br />
14 r<strong>and</strong>omized trials <strong>of</strong> stress<br />
ulcer prophylaxis in critically ill<br />
patients. (Tryba, 1995)<br />
6 (outcome VAP) <strong>and</strong> 7<br />
(outcome mortality) r<strong>and</strong>omized<br />
trials <strong>of</strong> stress ulcer prophylaxis<br />
in critically ill patients. (Cook,<br />
1995)<br />
14 r<strong>and</strong>omized trials <strong>of</strong> stress<br />
ulcer prophylaxis in critically ill<br />
patients. Endpoints were<br />
gastrointestinal bleeding <strong>and</strong><br />
pneumonia. (Tryba. 1991)<br />
Level 1A,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
201<br />
Pneumonia*<br />
(95% CI)<br />
ranitidine vs.<br />
sucralfate: 1.35<br />
(1.07-1.70)<br />
ranitidine vs.<br />
placebo: 0.98<br />
(0.56-1.72)<br />
sucralfate vs.<br />
placebo: 2.21<br />
(0.86-5.65)<br />
ranitidine vs.<br />
sucralfate: 1.18<br />
(0.92-1.51)<br />
sucralfate vs. H2antagonist:<br />
0.77<br />
(0.60-1.01)<br />
H2-antagonist vs.<br />
placebo: 1.25<br />
(0.78-2.00)<br />
sucralfate vs. H2antagonist/<br />
antacid: 0.67<br />
(p
Table 17.4.1. Studies <strong>of</strong> stress ulcer prophylaxis (cont.)<br />
9 r<strong>and</strong>omized trials <strong>of</strong> stress<br />
ulcer prophylaxis in critically ill<br />
patients. (Tryba, 1991)<br />
8 r<strong>and</strong>omized trials <strong>of</strong> stress<br />
ulcer prophylaxis in critically ill<br />
patients studying the rate <strong>of</strong><br />
pneumonia with different drug<br />
regimens. (Cook, 1991)<br />
Level<br />
1A,<br />
Level 1<br />
Level<br />
1A,<br />
Level 1<br />
* Point estimates reflect odds ratio or relative risk.<br />
References<br />
sucralfate vs. H2antagonist/<br />
antacid:<br />
0.48 (p
12. Slaughter S, Hayden MK, Nathan C, Hu TC, Rice T, Van Voorhis J, et al. A comparison <strong>of</strong><br />
the effect <strong>of</strong> universal use <strong>of</strong> gloves <strong>and</strong> gowns with that <strong>of</strong> glove use alone on acquisition<br />
<strong>of</strong> vancomycin-resistant enterococci in a medical intensive care unit. Ann Intern Med.<br />
1996; 125: 448-456.<br />
13. Ben-Menachem T, McCarthy BD, Fogel R, Schiffman RM, Patel RV, Zarowitz BJ, et al.<br />
Prophylaxis for stress-related gastrointestinal hemorrhage: a cost effectiveness analysis.<br />
Crit Care Med. 1996; 24: 338-345.<br />
14. Erstad BL, Camamo JM, Miller MJ, Webber AM, Fortune J. Impacting cost <strong>and</strong><br />
appropriateness <strong>of</strong> stress ulcer prophylaxis at a university medical center. Crit Care Med.<br />
1997; 25: 1678-1684.<br />
15. Prod'hom G, Leuenberger P, Koerfer J, Blum A, Chiolero R, Schaller MD, et al.<br />
Nosocomial pneumonia in mechanically ventilated patients receiving antacid, ranitidine, or<br />
sucralfate as prophylaxis for stress ulcer. A r<strong>and</strong>omized controlled trial. Ann Intern Med.<br />
1994; 120: 653-662.<br />
Final Comment to Chapter 17<br />
Ventilator-associated pneumonia is common, costly, <strong>and</strong> morbid. This chapter confirms<br />
that there are several low-risk interventions that carry the potential to reduce the frequency <strong>of</strong><br />
this complication. Further research will be needed to confirm the benefit <strong>of</strong> promising practices<br />
(eg, semi-recumbency or continuous aspiration <strong>of</strong> subglottic secretions) or fully allay concerns<br />
regarding practices that have potential for harm (eg, antibiotic resistance with selective<br />
decontamination).<br />
203
Section C. Surgery, Anesthesia, <strong>and</strong> Perioperative Medicine<br />
Chapter 18. Localizing Care to High-Volume Centers<br />
Chapter 19. Learning Curves for New Procedures – the Case <strong>of</strong> Laparoscopic<br />
Cholecystectomy<br />
Chapter 20. Prevention <strong>of</strong> Surgical Site Infections<br />
Subchapter 20.1 Prophylactic Antibiotics<br />
Subchapter 20.2. Perioperative Normothermia<br />
Subchapter 20.3. Supplemental Perioperative Oxygen<br />
Subchapter 20.4. Perioperative Glucose Control<br />
Chapter 21. Ultrasound Guidance <strong>of</strong> Central Vein Catheterization<br />
Chapter 22. The Retained Surgical Sponge<br />
Chapter 23: Pre-Anesthesia Checklists To Improve <strong>Patient</strong> <strong>Safety</strong><br />
Chapter 24. The Impact Of Intraoperative Monitoring On <strong>Patient</strong> <strong>Safety</strong><br />
Chapter 25. Beta-blockers <strong>and</strong> Reduction <strong>of</strong> Perioperative Cardiac Events<br />
204
Chapter 18. Localizing Care to High-Volume Centers<br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
The extent to which experience in caring for illness—as represented by a higher volume<br />
<strong>of</strong> cases—impacts outcomes has been well studied over the last 20 years. An extensive literature<br />
covering a broad range <strong>of</strong> conditions <strong>and</strong> procedures documents superior outcomes for hospitals<br />
<strong>and</strong> physicians with higher patient volumes. 1-6 Drawing on such evidence, various investigators<br />
have projected substantial reductions in mortality from regionalizing certain high-risk procedures<br />
with established volume-outcome relationships. 4-6<br />
When volume-outcomes relationships reflect a “practice makes perfect” effect, it may be<br />
reasonable to use volumes to assess quality <strong>of</strong> care. However, such relationships may also reflect<br />
“selective referral,” 7-10 when anecdotal knowledge <strong>of</strong> the superior quality <strong>of</strong> high volume centers<br />
exists in the community. 11 In such cases, direct measurements <strong>of</strong> processes or outcomes may<br />
represent more appropriate quality measures than volume alone. 12,13<br />
In an era <strong>of</strong> cost containment <strong>and</strong> a growing need for accountability for quality <strong>of</strong> care,<br />
any connection between site or physician-specific experience <strong>and</strong> patient outcomes has far<br />
reaching implications for patients, payers, <strong>and</strong> governmental agencies. 14 In fact, the Leapfrog<br />
Group (a consortium <strong>of</strong> major purchasers <strong>and</strong> purchasing coalitions) has made patient volume<br />
one <strong>of</strong> their criteria for quality, <strong>and</strong> has recently begun a project examining evidence-based<br />
referrals to high-volume centers (see also Chapter 55). 15<br />
The Institute <strong>of</strong> Medicine (IOM) recently sponsored a workshop to examine the evidence<br />
supporting this relationship, 16 part <strong>of</strong> which included a systematic review <strong>of</strong> investigations <strong>of</strong> the<br />
volume-outcome association. 17 This chapter summarizes the evidence supporting volumeoutcomes<br />
relationships, drawing heavily on the IOM’s systematic review <strong>and</strong> the workshop's<br />
findings.<br />
Practice Description<br />
The use <strong>of</strong> information regarding volume <strong>and</strong> its proven or potential relationship with<br />
better outcomes may result in several actions. Simply providing patients with volume data may<br />
result in preferential selection <strong>of</strong> high-volume centers or providers. <strong>Patient</strong>s might also be<br />
incentivized to choose high-volume centers (eg, through reduced co-payments). Alternatively,<br />
payers may elect to contract only with high-volume centers, or provide higher payments to these<br />
sites. Finally, low-volume centers (eg, a hospital failing to meet a minimum threshold <strong>of</strong> bypass<br />
operations) or providers (eg, a cardiologist failing to perform a minimum number <strong>of</strong><br />
angioplasties) might be restricted from continuing to perform the practice, through some<br />
combination <strong>of</strong> credentialing, accreditation, or regulatory actions (see Chapter 56).<br />
Opportunities for Impact<br />
Assuming that a feasible method could be developed to localize care to high-volume<br />
centers, a significant effect on patient safety <strong>and</strong> outcomes is likely. A recent study suggested<br />
that more than 500 deaths could be avoided annually in California alone if care for disorders<br />
with established volume-outcomes relationships were localized to more experienced centers.<br />
Extrapolating nationally, such localization would save 4000 lives. 6 In the case <strong>of</strong> acute<br />
myocardial infarction, transferring the care <strong>of</strong> these patients from hospitals in the lowest volume<br />
205
quartile to those in the highest would save 2.3 lives per 100 patients. 4 The corresponding<br />
“number needed to treat” (NNT) <strong>of</strong> 50 falls within the range <strong>of</strong> many accepted therapeutic<br />
interventions.<br />
Little information exists to assess differences in quality <strong>of</strong> care or rates <strong>of</strong> adverse events<br />
in high or low-volume sites. For example, the extent to which increasing volume leads to fewer<br />
medical errors or other direct impacts on patient safety (as opposed to specific improvement in<br />
care processes for discrete procedures, which would fall outside our definition <strong>of</strong> patient safety<br />
practices (Chapter 1)) is unknown. Ongoing prospective initiatives such as those proposed by the<br />
Leapfrog Group 15 may better quantify the various impacts <strong>of</strong> localizing care to high-volume<br />
centers.<br />
Study Designs<br />
We analyzed one large systematic review <strong>of</strong> 88 studies examining the relationship<br />
between volume <strong>and</strong> outcomes. 17 Using a structured MEDLINE search, this review included<br />
studies that examined health outcomes as the dependent variable with hospital <strong>and</strong>/or physician<br />
volume as an independent variable. The IOM review included medical <strong>and</strong> surgical conditions<br />
such as coronary artery bypass grafting, pediatric cardiac surgery, carotid endarterectomy,<br />
abdominal aortic aneurysm repair, cancer surgery, coronary angioplasty, acute myocardial<br />
infarction, AIDS, <strong>and</strong> multiple procedures. This chapter reviews the relationship for surgical<br />
illnesses only.<br />
The source studies <strong>of</strong> the IOM review were entirely observational in nature. Close<br />
attention was paid to risk adjustment <strong>and</strong> statistical methods in assessment <strong>of</strong> results, <strong>and</strong> criteria<br />
for inclusion in the review selected for population or community-based samples. Thus, the level<br />
<strong>of</strong> the design is classified as Level 3A.<br />
Study Outcomes<br />
The IOM systematic review examined health outcomes as related to hospital or physician<br />
volume. The primary outcome <strong>of</strong> interest was mortality. Other clinical outcomes were chosen<br />
based on complications specific to the surgical procedure (eg, stroke following carotid<br />
endarterectomy).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Results <strong>of</strong> the systematic review are outlined in Table 18.1. The studies reviewed were <strong>of</strong><br />
variable design <strong>and</strong> analytic sophistication, with more recent investigations generally being <strong>of</strong><br />
higher quality. For all procedures, there was a consistent trend toward an association between<br />
improved outcomes <strong>and</strong> higher hospital or physician-specific volume.* The evidence supporting<br />
the volume-outcomes relationship was similar when looking at hospital volume (78% <strong>of</strong> studies<br />
* This trend has one major exception, a study <strong>of</strong> volume-outcome relationships for 8 major<br />
surgical procedures at Veterans Affairs hospitals across the country (Khuri SF, et al. Relation <strong>of</strong><br />
surgical volume to outcome in eight common operations: results from the VA National Surgical<br />
Quality Improvement Program. Ann Surg. 1999;230:414-429). This comprehensive national<br />
study found no significant volume-outcome relationship for any <strong>of</strong> the 8<br />
procedures analyzed. While it is tempting to attribute these negative findings to unique aspects<br />
<strong>of</strong> the VA system, this is also one <strong>of</strong> the few studies to employ robust risk-adjustment using<br />
clinical <strong>and</strong> not administrative data. The authors <strong>of</strong> the IOM review take particular note <strong>of</strong> the<br />
superior methodologic features <strong>and</strong> negative findings <strong>of</strong> this study.<br />
206
showed an association) <strong>and</strong> physician volume (74% showed an association); the former was the<br />
more frequently analyzed variable.<br />
The impact <strong>of</strong> volume on outcomes varied across procedures <strong>and</strong> diagnoses. The effect<br />
was most marked for complex cancer surgeries (ie, esophagectomy, pancreatoduodenectomy),<br />
with numbers needed to treat (NNT) between 7 <strong>and</strong> 17. For commonly performed surgeries such<br />
as coronary artery bypass grafting, the NNT was generally larger, but still within the range <strong>of</strong><br />
other accepted therapies. Carotid endarterectomy appeared to have a much higher NNT, but this<br />
may be because the major adverse outcome <strong>of</strong> this surgery (ie, stroke) is <strong>of</strong>ten an indication for<br />
surgery as well as a complication <strong>of</strong> it. This relationship cannot be discerned from administrative<br />
data, <strong>and</strong> the studies upon which the NNT is based reported mortality, a less frequent<br />
complication <strong>of</strong> carotid surgery than stroke, as a primary outcome. The few studies that collected<br />
primary data (<strong>and</strong> would have been able to determine this important difference) were generally<br />
small <strong>and</strong> <strong>of</strong> lower quality design, making their findings suspect.<br />
The authors <strong>of</strong> the IOM systematic review conclude that the volume-outcomes<br />
relationship exists, but they raise some caveats when interpreting the literature as a whole. They<br />
first note that the literature describing the relationship in greatest detail come from a few single-<br />
State databases (63% <strong>of</strong> studies coming from New York State alone), possibly limiting the<br />
generalizability <strong>of</strong> their results. Although few studies employed rigorous risk adjustment<br />
methodologies using clinical data, the relationship between volume <strong>and</strong> outcomes was consistent<br />
in these studies. Also, they raise a note <strong>of</strong> caution in interpreting this relationship because <strong>of</strong> the<br />
possibility that publication bias (ie, studies that fail to demonstrate a volume-outcome<br />
relationship might be less likely to be submitted or accepted for publication) has affected this<br />
literature. Finally, they point out that the precise mechanism by which outcomes are improved<br />
has yet to be elucidated; no study has reported the independent effects <strong>of</strong> ancillary personnel<br />
expertise or hospital system factors on patient outcomes. For example, in interpreting the<br />
improved outcomes <strong>of</strong> high volume centers in coronary artery bypass grafting, we do not<br />
presently know what the relative contributions are <strong>of</strong> the surgeon, cardiac bypass team, cardiac<br />
anesthesiologist, <strong>and</strong> hospital resources (eg, dedicated cardiothoracic intensive care units).<br />
Potential for Harm<br />
In the summary accompanying the IOM workshop's proceedings, several potential<br />
pitfalls <strong>of</strong> localizing care to high volume settings were noted, as follows 16 :<br />
• The focus on high volume providers may be a “distracting priority,” <strong>and</strong><br />
similar improvements in care may be achieved through more traditional local<br />
quality improvement measures.<br />
• Hospitals with high volumes may use that data to misrepresent their<br />
experience in the absence <strong>of</strong> true outcomes data.<br />
• High volume centers may achieve too much contractual leverage, leading to<br />
price inflation.<br />
• Counting procedures may lead to perverse incentives to perform procedures<br />
that are not appropriate.<br />
• Requiring high volumes will impede entry <strong>of</strong> new competitors into the<br />
marketplace.<br />
• Narrowing the choice <strong>of</strong> providers may negatively impact patient satisfaction<br />
<strong>and</strong> override patients’ preferences for care (for example, if patients are forced<br />
to travel long distances to receive care at a high volume center).<br />
207
Several <strong>of</strong> these concerns have been borne out in published studies. In a study <strong>of</strong><br />
pediatric trauma centers, Tepas et al suggested that increases in volume may strain provider<br />
resources <strong>and</strong> worsen patient outcomes. 18 Even assuming potential improvements in patient<br />
outcomes, diverting patients to large referral centers has important health policy implications 14<br />
<strong>and</strong> may decrease patient satisfaction. 19<br />
Costs <strong>and</strong> Implementation<br />
The major barriers to implementing a selective referral program based on hospital<br />
volume include the potential harms listed above, as well as several additional factors. These may<br />
include patients' preference for care near home, lack <strong>of</strong> resources to travel, inability to transfer<br />
unstable patients to high-volume centers, loss <strong>of</strong> access to care in areas where low-volume<br />
services are discontinued (particularly rural areas), <strong>and</strong> resistance by providers to quality<br />
measurement activities. Finally, existing high volume centers may lack the capacity to accept<br />
additional patients. When they do not, further increases in volume could lead to increased rates<br />
<strong>of</strong> adverse events due to over-stressing the system, as was noted for pediatric trauma. 18<br />
Costs <strong>of</strong> this practice are not explicitly addressed in the IOM report, but widespread<br />
implementation <strong>of</strong> selective referrals would depend on the collection <strong>of</strong> detailed <strong>and</strong> accurate<br />
data (risk adjustment, process, <strong>and</strong> outcomes data), at substantial cost. Additionally, a<br />
nationwide systematic referral model may require augmenting capability <strong>of</strong> high-volume<br />
hospitals through additional construction or other major infrastructure investments. Finally, the<br />
travel <strong>and</strong> inconvenience costs <strong>of</strong> having patients obtain care at institutions outside the local<br />
community will be borne by either the system or the patients themselves—a cost which may be<br />
compounded further by patients’ <strong>and</strong> families’ lost productivity.<br />
Comment<br />
Changing practice patterns based on the compelling data linking volume <strong>and</strong> improved<br />
outcomes is a complex task involving patients, hospitals <strong>and</strong> communities, as well as payers <strong>and</strong><br />
employers. Actually closing hospitals or designating specific health care centers as “magnet”<br />
sites for care <strong>of</strong> specific illnesses would require, in addition to a wholesale change in health care<br />
systems, major infusions <strong>of</strong> resources <strong>and</strong> difficult political choices. For these reasons alone, this<br />
policy decision seems unlikely. Alternatively, hospitals or medical groups may use physicianspecific<br />
outcomes information to make decisions about staffing needs (eg, hiring a lower number<br />
<strong>of</strong> surgeons to ensure that all operators have high volumes), although this too seems unlikely<br />
unless meeting volume thresholds become m<strong>and</strong>atory or are strongly incentivized.<br />
The Leapfrog Group's efforts represent one <strong>of</strong> the first major initiatives to use empiric<br />
data to direct selective referrals using volume data. As an initiative begun <strong>and</strong> carried out within<br />
a specific employer/health care purchaser system it may be limited in its generalizability.<br />
However, prospective evaluation <strong>of</strong> the effort will yield important information regarding the<br />
costs <strong>and</strong> outcomes <strong>of</strong> such a referral system <strong>and</strong> its impact on patient satisfaction with care.<br />
Outside <strong>of</strong> the Leapfrog effort, the widespread publication <strong>of</strong> outcomes, especially<br />
mortality data, has been proposed as a way to help consumers to make more informed health care<br />
choices. A recent systematic review <strong>of</strong> this paradigm or “strategy” suggests that its impact has<br />
been mixed. 20 There are no data to suggest that patients or payers will make major changes in<br />
health care purchasing decisions when provided with volume data alone. To date, press reports<br />
<strong>of</strong> particularly noteworthy errors seem to have more <strong>of</strong> an impact on patients' choices <strong>of</strong> care<br />
than information about volume or even outcomes <strong>of</strong> care. 20,21<br />
208
In addition to the potential use <strong>of</strong> volume data to guide health care purchasing decisions,<br />
the IOM workshop <strong>and</strong> authors <strong>of</strong> the systematic review recommend using volume as one <strong>of</strong><br />
several quality measures to initiate local or regional quality improvement efforts. This data may<br />
motivate or inform care improvement efforts at low-volume sites (or those with worse-thanexpected<br />
outcomes) through use <strong>of</strong> site visits, feedback <strong>of</strong> risk-adjusted outcomes information,<br />
<strong>and</strong> assessment <strong>of</strong> care processes. This collaborative approach to quality improvement has been<br />
used successfully in the VA National Surgical Quality Improvement Project 22 <strong>and</strong> in several<br />
projects in New York State. 23,24 However, it seems likely that as the volume-outcome<br />
relationship becomes better defined <strong>and</strong> understood, its influence on the health care choices <strong>of</strong><br />
both patients <strong>and</strong> payers is likely to grow.<br />
209
Table 18.1. Summary <strong>of</strong> findings from IOM systematic review 17 <strong>of</strong> volume-outcome<br />
relationship*<br />
Condition Number <strong>of</strong><br />
Studies<br />
Reviewed<br />
Coronary<br />
artery bypass<br />
grafting<br />
9<br />
Carotid<br />
endarterectomy 19<br />
Cancer surgery 20<br />
Abdominal<br />
aortic<br />
aneurysm<br />
repair<br />
Pediatric<br />
cardiac<br />
surgery<br />
12<br />
3<br />
Comments<br />
All studies used appropriate risk adjustment<br />
VOR for both physicians <strong>and</strong> hospital: 7 studies<br />
Absolute difference in mortality between high- <strong>and</strong> low-volume<br />
centers <strong>and</strong> surgeons was 3-10% (NNT 10-33)<br />
Some evidence to suggest that centers <strong>and</strong> surgeons with good<br />
outcomes experienced increasing volumes over time<br />
(“selective referrals”)<br />
Only 9 studies performed adequate risk adjustment<br />
Most studies employed administrative data, making accurate<br />
ascertainment <strong>of</strong> postoperative stroke impossible<br />
VOR found for surgeon: 9 studies<br />
VOR found for hospital: 7 studies<br />
Absolute difference in mortality between high- <strong>and</strong> low-volume<br />
hospital/surgeon was 0.2-0.9% (NNT 100-500)<br />
Risk adjustment methods variable, most are dependent on<br />
administrative data only<br />
VOR most marked for rare cancers/procedures such as pancreatic<br />
resection <strong>and</strong> esophageal surgery<br />
VOR unclear for common surgery such as colectomy <strong>and</strong><br />
pneumonectomy<br />
For esophagectomy, absolute difference in mortality between<br />
high- <strong>and</strong> low-volume hospitals was 11-13.9% (NNT 7-9)<br />
For pancreatic resection, difference in mortality between high- <strong>and</strong><br />
low-volume hospitals gives NNT 10-15<br />
11 studies performed adequate risk adjustment<br />
VOR for hospitals <strong>and</strong> physicians noted<br />
Absolute reduction in mortality due to surgeon or hospital volume<br />
was 5-9% (NNT 11-20)<br />
“Selective referrals” noted<br />
All studies used appropriate risk adjustment<br />
VOR found for both hospital <strong>and</strong> surgeon volume<br />
Absolute difference in mortality due to surgeon or hospital volume<br />
was 3% (NNT 33)<br />
Possibly greater benefit for more complex/sicker patients<br />
* NNT indicates number needed to treat; VOR, volume-outcome relationship<br />
210
References<br />
1. Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical<br />
relation between surgical volume <strong>and</strong> mortality. N Engl J Med. 1979;301:1364-1369.<br />
2. Begg CB, Cramer LD, Hoskins WJ, Brennan MF. Impact <strong>of</strong> hospital volume on operative<br />
mortality for major cancer surgery. JAMA. 1998;280:1747-1751.<br />
3. Glasgow RE, Showstack JA, Katz PP, Corvera CU, Warren RS, Mulvihill SJ. The<br />
relationship between hospital volume <strong>and</strong> outcomes <strong>of</strong> hepatic resection for hepatocellular<br />
carcinoma. Arch Surg. 1999;134:30-35.<br />
4. Thiemann DR, Coresh J, Oetgen WJ, Powe NR. The association between hospital volume<br />
<strong>and</strong> survival after acute myocardial infarction in elderly patients. N Engl J Med.<br />
1999;340:1640-1648.<br />
5. Birkmeyer JD, Lucas FL, Wennberg DE. Potential benefits <strong>of</strong> regionalizing major surgery<br />
in Medicare patients. Eff Clin Pract. 1999;2:277-283.<br />
6. Dudley RA, Johansen KL, Br<strong>and</strong> R, Rennie DJ, Milstein A. Selective referral to highvolume<br />
hospitals: estimating potentially avoidable deaths. JAMA. 2000;283:1159-1166.<br />
7. Luft HS, Hunt SS, Maerki SC. The volume-outcome relationship: practice-makes-perfect<br />
or selective- referral patterns? Health Serv Res. 1987;22:157-182.<br />
8. Hughes RG, Garnick DW, Luft HS, McPhee SJ, Hunt SS. Hospital volume <strong>and</strong> patient<br />
outcomes. The case <strong>of</strong> hip fracture patients. Med Care. 1988;26:1057-1067.<br />
9. Farley DE, Ozminkowski RJ. Volume-outcome relationships <strong>and</strong> in-hospital mortality: the<br />
effect <strong>of</strong> changes in volume over time. Med Care. 1992;30:77-94.<br />
10. Hamilton BH, Ho V. Does practice make perfect? Examining the relationship between<br />
hospital surgical volume <strong>and</strong> outcomes for hip fracture patients in Quebec. Med Care.<br />
1998;36:892-903.<br />
11. Luft HS, Garnick DW, Mark DH, et al. Does quality influence choice <strong>of</strong> hospital? JAMA.<br />
1990;263:2899-2906.<br />
12. Flood AB, Scott WR, Ewy W. Does practice make perfect? Part I: The relation between<br />
hospital volume <strong>and</strong> outcomes for selected diagnostic categories. Med Care. 1984;22:98-<br />
114.<br />
13. Luft HS, Hunt SS. Evaluating individual hospital quality through outcome statistics. JAMA.<br />
1986;255:2780-2784.<br />
14. Phillips KA, Luft HS. The policy implications <strong>of</strong> using hospital <strong>and</strong> physician volumes as<br />
"indicators" <strong>of</strong> quality <strong>of</strong> care in a changing health care environment. Int J Qual Health<br />
Care. 1997;9:341-348.<br />
15. The Leapfrog Group for patient safety: Rewarding higher st<strong>and</strong>ards. In: Group TL, ed.<br />
www.leapfrog.org. 2001.<br />
16. Hewitt M. (National Academy <strong>of</strong> sciences). Interpreting the volume-outcome relationship<br />
in the context <strong>of</strong> health care quality. 2000.<br />
17. Halm EA, Lee C, Chassin M. (Institute <strong>of</strong> Medicine). How is volume related to quality in<br />
health care?: A systematic review <strong>of</strong> the research literature. 2000.<br />
18. Tepas JJ, 3rd, Patel JC, DiScala C, Wears RL, Veldenz HC. Relationship <strong>of</strong> trauma patient<br />
volume to outcome experience: can a relationship be defined? J Trauma. 1998;44:827-830.<br />
19. Finlayson SR, Birkmeyer JD, Tosteson AN, Nease RF, Jr. <strong>Patient</strong> preferences for location<br />
<strong>of</strong> care: implications for regionalization. Med Care. 1999;37:204-209.<br />
20. Marshall MN, Shekelle PG, Leatherman S, Brook RH. The public release <strong>of</strong> performance<br />
data: what do we expect to gain? A review <strong>of</strong> the evidence JAMA. 2000;283:1866-1874.<br />
211
21. Mennemeyer ST, Morrisey MA, Howard LZ. Death <strong>and</strong> reputation: how consumers acted<br />
upon HCFA mortality information. Inquiry. 1997;34:117-128.<br />
22. Khuri SF, Daley J, Henderson W, et al. The Department <strong>of</strong> Veterans Affairs' NSQIP: the<br />
first national, validated, outcome-based, risk-adjusted, <strong>and</strong> peer-controlled program for the<br />
measurement <strong>and</strong> enhancement <strong>of</strong> the quality <strong>of</strong> surgical care. National VA Surgical<br />
Quality Improvement Program. Ann Surg. 1998;228:491-507.<br />
23. Chassin MR, Hannan EL, DeBuono BA. Benefits <strong>and</strong> hazards <strong>of</strong> reporting medical<br />
outcomes publicly. N Engl J Med. 1996;334:394-398.<br />
24. Coronary Artery Bypass Surgery in New York State, 1992-1994 Albany, NY: New York<br />
State Department <strong>of</strong> Health. 1996.<br />
212
Chapter 19. Learning Curves for New Procedures – the Case <strong>of</strong><br />
Laparoscopic Cholecystectomy<br />
Verna C. Gibbs, MD<br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Minimal access surgery began in the early 1980s with the introduction <strong>of</strong> laparoscopic<br />
fallopian tube ligation. The first laparoscopic cholecystectomy was performed 7 years later, <strong>and</strong><br />
was rapidly embraced as the preferred method for cholecystectomy despite a lack <strong>of</strong> evidence to<br />
support the safety <strong>of</strong> the new technique. 1-6 In response to several deaths <strong>and</strong> complications<br />
associated with laparoscopic cholecystectomy, the New York State Department <strong>of</strong> Health issued<br />
guidelines for the credentialing <strong>of</strong> surgeons who wished to perform the procedure. 7,8 At the same<br />
time, a National Institutes <strong>of</strong> Health Consensus conference published recommendations<br />
regarding indications for laparoscopic cholecystectomy. 1<br />
Clinical trials comparing the laparoscopic procedure with other approaches eventually<br />
revealed that the newer procedure to be less morbid than traditional open cholecystectomy, or<br />
even mini-laparotomy. 9-16 Importantly, though, it also became clear that acquiring the skills to<br />
perform this new procedure involved a substantial surgical “learning curve.” 17-22 This learning<br />
curve no longer affects patients undergoing laparoscopic cholecystectomy, as training has<br />
become a required part <strong>of</strong> all surgical residency programs, with graduating surgery residents<br />
typically having performed more than 50 laparoscopic cholecystectomies. 23,24 However, the<br />
growth <strong>of</strong> laparoscopic cholecystectomy has been followed by equally rapid development <strong>and</strong><br />
application <strong>of</strong> minimal access procedures in virtually every surgical specialty. This chapter<br />
considers the patient safety issues that arise with the diffusion <strong>of</strong> a new procedurally-based<br />
technology. It highlights the “learning curve” inherent in any new procedure, as competence<br />
invariably grows with practice (see Chapter 18). Because it is so widely performed <strong>and</strong> has the<br />
largest literature describing its record, this chapter focuses on lessons learned from the<br />
introduction <strong>of</strong> laparoscopic cholecystectomy.<br />
Practice Description<br />
Although there are a wide variety <strong>of</strong> procedure-specific techniques in minimal access<br />
surgery, all operations utilize videoscopic or digital imaging in conjunction with remote<br />
manipulation. In general, the operator works from two-dimensional, magnified video images <strong>of</strong><br />
the operative field while manipulating long narrow instruments placed into the operating cavity<br />
from outside the body. To become pr<strong>of</strong>icient in minimal access techniques, the surgeon must<br />
develop skills in interpreting a three-dimensional environment as a two-dimensional image, <strong>and</strong><br />
learn how to do familiar tasks (eg, suture) with familiar instruments in an unfamiliar manner.<br />
Importantly, the surgeon never actually touches the tissue being moved with his or her h<strong>and</strong>s. 21,25<br />
This loss <strong>of</strong> tactile input is the major factor in making minimal access techniques difficult to<br />
learn.<br />
213
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Complications from minimal access procedures fall into 2 general categories: those<br />
directly resulting from the laparoscopic intervention (ie, trocar injuries, dissection injuries,<br />
insufflation-associated events) 26 <strong>and</strong> those associated with the operation itself (eg, bile duct<br />
injury with cholecystectomy, gastric injury with fundoplication).* 17-22 For laparoscopic<br />
cholecystectomy, a survey <strong>of</strong> 1750 surgery department chairpersons reported that bowel <strong>and</strong><br />
vascular injuries (laparoscopic-specific injury) occurred in 0.14% <strong>and</strong> 0.25% <strong>of</strong> cases<br />
respectively, while the rate <strong>of</strong> bile duct injuries (procedure-specific) was 0.6%. 3 The bile duct<br />
injury rate was lower at institutions that had performed more than 100 cases. Although this study<br />
was large, as a survey it likely underestimates the true rate <strong>of</strong> complications. Other multiinstitutional<br />
reports <strong>of</strong> laparoscopic cholecystectomy in the United States have noted a similar<br />
bile duct injury rate <strong>and</strong> report mortality rates <strong>of</strong> 0.04% to 0.1%. 5,21,25,29 Although undoubtedly<br />
underestimated, even these rates are higher than those associated with open cholecystectomy.<br />
Opportunities for Impact<br />
As set forth above, laparoscopic cholecystectomies is a st<strong>and</strong>ard part <strong>of</strong> surgical<br />
residency training. 23,24 Nearly all general surgeons who did not train in the current era have<br />
received postgraduate training in basic laparoscopic techniques. This training <strong>of</strong>ten takes the<br />
form <strong>of</strong> a 2-3 day course involving h<strong>and</strong>s-on experience with animal models <strong>and</strong> laboratoryskills<br />
sessions, then observation or assisting with cases primarily performed by another surgeon,<br />
finally, performing cases supervised by an experienced laparoscopist, <strong>and</strong> finally performing<br />
cases independently. 6,24,30<br />
Pr<strong>of</strong>iciency in traditional techniques does not automatically translate to minimal access<br />
methods. We identified one study investigating whether the level <strong>of</strong> surgical training (attending<br />
surgeon versus chief resident) affected the performance <strong>of</strong> cholecystectomy. 17 This study<br />
suggested that, despite operating on more complex patients, the adverse event rate for surgical<br />
chief residents learning laparoscopic cholecystectomy was similar to that <strong>of</strong> attending physicians<br />
(who had far more operative experience).<br />
Similarly, skills obtained with one minimal access procedure do not transfer to others.<br />
The operation <strong>of</strong> laparoscopic fundoplication for gastroesophageal reflux disease entered into<br />
surgical practice after laparoscopic cholecystectomy. Therefore, most surgeons who performed it<br />
had basic laparoscopic skill sets. Studies investigating its implementation suggest a threshold <strong>of</strong><br />
25-30 cases before surgeons attain pr<strong>of</strong>iciency, 31-34 in contrast to a threshold <strong>of</strong> about 15-20<br />
cases for laparoscopic cholecystectomy. 25,29,30,35 In recognition <strong>of</strong> this lack <strong>of</strong> transferability,<br />
performing laparoscopic cholecystectomies does not qualify a surgeon to perform laparoscopic<br />
colon resection without first undergoing additional training specifically in laparoscopic colon<br />
techniques. 6,36 The need for specific new training has been seen with other procedures as well. 37-<br />
41<br />
* A third problem concerns the potential adverse effects at the population. While the procedure<br />
itself may be less morbid, the rate at which this ‘safer’ procedure is performed may increase<br />
substantially to the point that the complications experienced by the patient population as a whole<br />
do not decrease. 27 This problem has been documented with laparoscopic cholecystectomy, 12,28<br />
but processes for improved patient selection have not received sufficient testing to permit review<br />
<strong>of</strong> a particular safety practice relating to procedure “appropriateness.”<br />
214
Thus, determination <strong>of</strong> a minimum number <strong>of</strong> cases required to become pr<strong>of</strong>icient (as an<br />
aid to credentialing or proctoring activities), or determining methods to shorten the time needed<br />
to attain this level <strong>of</strong> skill will have a clear impact on patients undergoing minimal access<br />
procedures. This impact is likely to become greater as the breadth <strong>and</strong> complexity <strong>of</strong> procedures<br />
approached through minimal access techniques grows.<br />
Study Designs<br />
Using a structured MEDLINE search, we sought to identify papers that either reported a<br />
threshold number <strong>of</strong> cases required to attain pr<strong>of</strong>iciency, or explicitly compared training<br />
techniques for laparoscopic cholecystectomy in terms <strong>of</strong> patient outcomes. Our literature search<br />
found a large number <strong>of</strong> reports describing retrospective analyses or reports from procedure<br />
registries. These studies did not report training protocols, specific physician characteristics or<br />
surgeon-specific volumes, <strong>and</strong> were therefore excluded from our review. In addition, we found<br />
several review articles. 42-44 The first <strong>of</strong> these reviews focused primarily on statistical techniques<br />
for documenting <strong>and</strong> analyzing learning curves, 42 <strong>and</strong> the other two 43,44 did not meet criteria for<br />
systematic reviews. Nonetheless, the bibliographies <strong>of</strong> all 3 articles were scanned for relevant<br />
studies.<br />
Study Outcomes<br />
All studies reported clinical outcomes (Level 1), most commonly bile duct injury. Other<br />
complications (eg, insufflation or trocar injuries) were very unusual, <strong>and</strong> mortality was rare.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Three studies specifically addressed the relationship between surgical experience in the<br />
performance <strong>of</strong> laparoscopic cholecystectomy <strong>and</strong> a well-defined adverse outcome, bile duct<br />
injury. The first reported 15 bile duct injuries in over 8000 laparoscopic cholecystectomies. 30<br />
Multivariate analysis showed that the only significant factor in predicting this adverse event was<br />
the experience <strong>of</strong> the surgeon (p=0.001). Rapidity <strong>of</strong> learning was not significantly related to a<br />
surgeon's age, size <strong>of</strong> practice, or hospital setting. Ninety percent <strong>of</strong> injuries were predicted to<br />
occur during a surgeon’s first 30 cases. Gigot et al 35 reported the incidence <strong>of</strong> bile duct injury<br />
was 1.3% when the surgeon had performed fewer than 50 cases <strong>and</strong> 0.35% afterwards (p100 cases. 35<br />
Similar results have been observed in other studies <strong>of</strong> bile duct injury with laparoscopic<br />
cholecystectomy, 45 suggesting that something beyond the learning curve accounts for many<br />
laparoscopic errors. Examination from error analysis suggests that these events occur because <strong>of</strong><br />
visual <strong>and</strong> perceptual difficulties involved in the operator/equipment interface. 46,47<br />
We identified one prospective cohort study from Japan that compared the effectiveness <strong>of</strong><br />
2 training courses for laparoscopic cholecystectomy. 2 In this study, one group <strong>of</strong> 8 surgeons was<br />
assigned to 10 supervised laparoscopic cholecystectomies as part <strong>of</strong> their initial training; a<br />
second group <strong>of</strong> 8 surgeons had only 2 observed training sessions. Complications that occurred<br />
over 21 months after completion <strong>of</strong> the 4 months <strong>of</strong> training were assessed by means <strong>of</strong> a<br />
questionnaire sent to all participants. The surgeons who had trained with 10 supervised<br />
procedures had 0.5% major complications (bleeding, bile duct injury or bile leakage) versus 2%<br />
for the surgeons trained with only 2 procedures (p=0.03). The incidence <strong>of</strong> major complications<br />
occurring during the initial 10 operations was higher among unsupervised surgeons (p=0.005).<br />
Outside <strong>of</strong> laparoscopic cholecystectomy, we identified one study examining<br />
complication rates <strong>of</strong> urologic laparoscopic surgeons after completing a 2-day training seminar. 48<br />
215
Because the study relied on surveys <strong>of</strong> participants at 3 <strong>and</strong> 12 months after course completion, it<br />
is likely biased toward underestimating the incidence <strong>of</strong> adverse events. Nevertheless the study<br />
showed that surgeons who performed procedures without additional training were 3 times more<br />
likely to have at least one complication compared with surgeons who sought additional training.<br />
Additionally the presence <strong>of</strong> skilled associates <strong>and</strong> the development <strong>of</strong> a surgical team impact<br />
favorably on reducing the number <strong>of</strong> complications. 48,49<br />
Potential for Harm<br />
It is unlikely that the specific introduction <strong>of</strong> training <strong>and</strong> requirements for a baseline <strong>of</strong><br />
surgical experience in minimal access surgery procedures will lead to adverse events.<br />
Costs <strong>and</strong> Implementation<br />
In the initial introduction <strong>of</strong> laparoscopic cholecystectomy most surgeons attended at<br />
least one didactic <strong>and</strong> h<strong>and</strong>s-on course in operative technique <strong>and</strong> instrument manipulation.<br />
Some <strong>of</strong> these experiences included operative procedures in animals <strong>and</strong> utilization <strong>of</strong> static<br />
training boxes <strong>and</strong> devices that may speed the acquisition <strong>of</strong> perceptual <strong>and</strong> motor skills. The<br />
need for practice in the development <strong>of</strong> technical skills is vital, <strong>and</strong> laboratory courses, training<br />
devices <strong>and</strong> simulators (Chapter 45) are being tested for their ability to improve operator<br />
skills. 50-55 There is insufficient evidence to make a recommendation <strong>of</strong> the superiority <strong>of</strong> one<br />
training method over any other <strong>and</strong> costs are presently undetermined.<br />
After sufficient laboratory experience <strong>and</strong> practice, it is recommended that the procedure<br />
be performed initially in only carefully selected patients under the supervision <strong>of</strong> an experienced<br />
surgeon. 56-58 How practicing groups <strong>of</strong> providers obtain an experienced consultant is<br />
variable. 59,60 Surgical training fellowships in minimal access surgery have been established <strong>and</strong><br />
graduates <strong>of</strong> these programs have the requisite expertise to supervise others. Unfortunately, such<br />
fellowship graduates are in short supply. Telemedicine may play a role in the mentoring <strong>and</strong><br />
technical support for surgeons performing new techniques as it provides a means for remote<br />
performance <strong>of</strong> a procedure with real-time expert supervision <strong>and</strong> guidance. 61,62<br />
Comment<br />
Minimal access surgery has become extraordinarily popular largely in response to market<br />
forces. The single most important predictor <strong>of</strong> adverse events in minimal access procedures is<br />
the experience <strong>of</strong> the provider with the specific operation. Surgeons must acquire the necessary<br />
technical skills <strong>and</strong> expertise before performing new procedures on patients. Hospitals <strong>and</strong><br />
payers should help ensure that providers possess the requisite experience before allowing<br />
procedures to be performed in their facilities or paying for them, since patients alone will<br />
generally be unable to determine surgeon competency.<br />
A number <strong>of</strong> governing bodies <strong>and</strong> surgical societies have published guidelines that<br />
outline st<strong>and</strong>ards for training for postgraduate surgeons for skill acquisition in minimal access<br />
surgery, 1,56-58,63-66 but these recommendations are based more on common sense <strong>and</strong> clinical<br />
experience than rigorous evidence. It is not known how influential these guidelines are in the<br />
granting <strong>of</strong> privileges. Continued research is needed to determine the threshold for safe<br />
performance <strong>of</strong> this <strong>and</strong> other procedures, the most effective training methods to ensure<br />
competence, <strong>and</strong> strategies to minimize patient harm while proceduralists gain the experience<br />
they need to be competent <strong>and</strong> to train others.<br />
216
References<br />
1. Gallstones <strong>and</strong> laparoscopic cholecystectomy. NIH Consensus Development Panel on<br />
Gallstones <strong>and</strong> Laparoscopic Cholecystectomy. Surg Endosc. 1993; 7: 271-279.<br />
2. Yamashita Y, Kurohiji T, Kakegawa T. Evaluation <strong>of</strong> two training programs for<br />
laparoscopic cholecystectomy: incidence <strong>of</strong> major complications. World J Surg.<br />
1994;18:279-285.<br />
3. Deziel DJ, Millikan KW, Economou SG, Doolas A, Ko ST, Airan MC. Complications <strong>of</strong><br />
laparoscopic cholecystectomy: a national survey <strong>of</strong> 4,292 hospitals <strong>and</strong> an analysis <strong>of</strong><br />
77,604 cases. Am J Surg. 1993;165:9-14.<br />
4. Hodgson WJ, Byrne DW, Savino JA, Liberis G. Laparoscopic cholecystectomy. The early<br />
experience <strong>of</strong> surgical attendings compared with that <strong>of</strong> residents trained by apprenticeship.<br />
Surg Endosc. 1994; 8:1058-1062.<br />
5. Wherry DC, Rob CG, Marohn MR, Rich NM. An external audit <strong>of</strong> laparoscopic<br />
cholecystectomy performed in medical treatment facilities <strong>of</strong> the Department <strong>of</strong> Defense.<br />
Ann Surg. 1994; 220:626-634.<br />
6. Grundfest WS. Credentialing in an era <strong>of</strong> change. JAMA. 1993; 270: 2725.<br />
7. NY State issues guidelines for lap cholecystectomy. New York State Health Department.<br />
OR Manager. 1992;8:1,8-9.<br />
8. Green FL. New York State Health Department ruling–a "wake-up call" for all. Surg<br />
Endosc. 1992;6:271.<br />
9. Shea JA, Healey MJ, Berlin JA, Clarke JR, Malet PF, Staroscik RN, et al. Mortality <strong>and</strong><br />
complications associated with laparoscopic cholecystectomy. A meta-analysis. Ann Surg.<br />
1996;224:609-620.<br />
10. Majeed AW, Troy G, Nicholl JP, Smythe A, Reed MW, Stoddard CJ, et al. R<strong>and</strong>omised,<br />
prospective, single-blind comparison <strong>of</strong> laparoscopic versus small-incision<br />
cholecystectomy. Lancet. 1996;347:989-994.<br />
11. Kane RL, Lurie N, Borbas C, Morris N, Flood S, McLaughlin B, et al. The outcomes <strong>of</strong><br />
elective laparoscopic <strong>and</strong> open cholecystectomies. J Am Coll Surg. 1995;180:136-145.<br />
12. Steiner CA, Bass EB, Talamini MA, Pitt HA, Steinberg EP. Surgical rates <strong>and</strong> operative<br />
mortality for open <strong>and</strong> laparoscopic cholecystectomy in Maryl<strong>and</strong>. N Engl J Med.<br />
1994;330:403-408.<br />
13. McMahon AJ, Russell IT, Baxter JN, Ross S, Anderson JR, Morran CG, et al.<br />
Laparoscopic versus minilaparotomy cholecystectomy: a r<strong>and</strong>omised trial. Lancet.<br />
1994;343:135-138.<br />
14. Barkun JS, Barkun AN, Meakins JL. Laparoscopic versus open cholecystectomy: the<br />
Canadian experience. The McGill Gallstone Treatment Group. Am J Surg. 1993;165:455-<br />
458.<br />
15. Orl<strong>and</strong>o R, Russell JC, Lynch J, Mattie A. Laparoscopic cholecystectomy. A statewide<br />
experience. The Connecticut Laparoscopic Cholecystectomy Registry. Arch Surg.<br />
1993;128:494-8.<br />
16. Trondsen E, Reiertsen O, Andersen OK, Kjaersgaard P. Laparoscopic <strong>and</strong> open<br />
cholecystectomy. A prospective, r<strong>and</strong>omized study. Eur J Surg. 1993;159:217-221.<br />
17. Lekawa M, Shapiro SJ, Gordon LA, Rothbart J, Hiatt JR. The laparoscopic learning curve.<br />
Surg Laparosc Endosc. 1995;5:455-458.<br />
217
18. Dashow L, Friedman I, Kempner R, Rudick J, McSherry C. Initial experience with<br />
laparoscopic cholecystectomy at the Beth Israel <strong>Medical</strong> Center. Surg Gynecol Obstet.<br />
1992;175:25-30.<br />
19. Gilchrist BF, Vlessis AA, Kay GA, Swartz K, Dennis D. Open versus laparoscopic<br />
cholecystectomy: an initial analysis. J Laparoendosc Surg. 1991;1:193-196.<br />
20. Jones RM, Fletcher DR, MacLellan DG, Lowe AW, Hardy KJ. Laparoscopic<br />
cholecystectomy: initial experience. Aust N Z J Surg. 1991;61:261-266.<br />
21. Peters JH, Ellison EC, Innes JT, Liss JL, Nichols KE, Lomano JM, et al. <strong>Safety</strong> <strong>and</strong><br />
efficacy <strong>of</strong> laparoscopic cholecystectomy. A prospective analysis <strong>of</strong> 100 initial patients.<br />
Ann Surg. 1991;213:3-12.<br />
22. Sariego J, Spitzer L, Matsumoto T. The "learning curve" in the performance <strong>of</strong><br />
laparoscopic cholecystectomy. Int Surg. 1993;78:1-3.<br />
23. Dent TL. Training <strong>and</strong> privileging for new procedures. Surg Clin North Am. 1996;76:615-<br />
621.<br />
24. Parsa CJ, Organ CH, Jr., Barkan H. Changing patterns <strong>of</strong> resident operative experience<br />
from 1990 to 1997. Arch Surg. 2000;135:570-3.<br />
25. Zucker KA, Bailey RW, Gadacz TR, Imbembo AL. Laparoscopic guided cholecystectomy.<br />
Am J Surg. 1991;161:36-42.<br />
26. Philips PA, Amaral JF. Abdominal access complications in laparoscopic surgery. J Am<br />
Coll Surg. 2001;192:525-536.<br />
27. Roos NP, Black CD, Roos LL, Tate RB, Carriere KC. A population-based approach to<br />
monitoring adverse outcomes <strong>of</strong> medical care. Med Care. 1995;33:127-138.<br />
28. Escarce JJ, Chen W, Schwartz JS. Falling cholecystectomy thresholds since the<br />
introduction <strong>of</strong> laparoscopic cholecystectomy. JAMA. 1995;273:1581-1585.<br />
29. A prospective analysis <strong>of</strong> 1518 laparoscopic cholecystectomies. The Southern Surgeons<br />
Club. N Engl J Med. 1991;324:1073-1078.<br />
30. Moore MJ, Bennett CL. The learning curve for laparoscopic cholecystectomy. The<br />
Southern Surgeons Club. Am J Surg. 1995;170:55-59.<br />
31. Champault GG, Barrat C, Rozon RC, Rizk N, Catheline JM. The effect <strong>of</strong> the learning<br />
curve on the outcome <strong>of</strong> laparoscopic treatment for gastroesophageal reflux. Surg Laparosc<br />
Endosc Percutan Tech. 1999;9:375-381.<br />
32. Watson DI, Baigrie RJ, Jamieson GG. A learning curve for laparoscopic fundoplication.<br />
Definable, avoidable, or a waste <strong>of</strong> time? Ann Surg. 1996;224:198-203.<br />
33. Soot SJ, Eshraghi N, Farahm<strong>and</strong> M, Sheppard BC, Deveney CW. Transition from open to<br />
laparoscopic fundoplication: the learning curve. Arch Surg. 1999;134:278-281.<br />
34. Deschamps C, Allen MS, Trastek VF, Johnson JO, Pairolero PC. Early experience <strong>and</strong><br />
learning curve associated with laparoscopic Nissen fundoplication. J Thorac Cardiovasc<br />
Surg. 1998;115:281-288.<br />
35. Gigot J, Etienne J, Aerts R, Wibin E, Dallemagne B, Deweer F, et al. The dramatic reality<br />
<strong>of</strong> biliary tract injury during laparoscopic cholecystectomy. An anonymous multicenter<br />
Belgian survey <strong>of</strong> 65 patients. Surg Endosc. 1997;11:1171-1178.<br />
36. Agachan F, Joo JS, Weiss EG, Wexner SD. Intraoperative laparoscopic complications. Are<br />
we getting better? Dis Colon Rectum. 1996;39:S14-19.<br />
37. Hunter JG. Advanced laparoscopic surgery. Am J Surg. 1997;173:14-18.<br />
38. Janetschek G, Hobisch A, Holtl L, Bartsch G. Retroperitoneal lymphadenectomy for<br />
clinical stage I nonseminomatous testicular tumor: laparoscopy versus open surgery <strong>and</strong><br />
impact <strong>of</strong> learning curve. J Urol. 1996;156:89-93.<br />
218
39. Crawford DL, Phillips EH. Laparoscopic repair <strong>and</strong> groin hernia surgery. Surg Clin North<br />
Am. 1998;78:1047-1062.<br />
40. Meinke AK, Kossuth T. What is the learning curve for laparoscopic appendectomy? Surg<br />
Endosc. 1994;8:371-375.<br />
41. Perino A, Cucinella G, Venezia R, Castelli A, Cittadini E. Total laparoscopic hysterectomy<br />
versus total abdominal hysterectomy: an assessment <strong>of</strong> the learning curve in a prospective<br />
r<strong>and</strong>omized study. Hum Reprod. 1999;14:2996-2999.<br />
42. Ramsay CR, Grant AM, Wallace SA, Garthwaite PH, Monk AF, Russell IT. Assessment <strong>of</strong><br />
the learning curve in health technologies. A systematic review. Int J Technol Assess Health<br />
Care. 2000;16:1095-1108.<br />
43. Rogers DA, Elstein AS, Bordage G. Improving continuing medical education for surgical<br />
techniques: applying the lessons learned in the first decade <strong>of</strong> minimal access surgery. Ann<br />
Surg. 2001;233:159-166.<br />
44. Hamdorf JM, Hall JC. Acquiring surgical skills. Br J Surg. 2000;87:28-37.<br />
45. Steele RJ, Marshall K, Lang M, Doran J. Introduction <strong>of</strong> laparoscopic cholecystectomy in a<br />
large teaching hospital: independent audit <strong>of</strong> the first 3 years. Br J Surg. 1995;82:968-971.<br />
46. Olsen DO. Bile duct injuries during laparoscopic cholecystectomy: a decade <strong>of</strong> experience.<br />
J Hepatobiliary Pancreat Surg. 2000;7:35-39.<br />
47. Stewart L, Way LW. Bile duct injuries during laparoscopic cholecystectomy. Factors that<br />
influence the results <strong>of</strong> treatment. Arch Surg. 1995;130:1123-1128.<br />
48. See WA, Cooper CS, Fisher RJ. Predictors <strong>of</strong> laparoscopic complications after formal<br />
training in laparoscopic surgery. JAMA. 1993;270:2689-2692.<br />
49. Sexton JB, Thomas EJ, Helmreich RL. Error, stress, <strong>and</strong> teamwork in medicine <strong>and</strong><br />
aviation: cross sectional surveys. BMJ. 2000;320:745-749.<br />
50. Noar MD. The next generation <strong>of</strong> endoscopy simulation: minimally invasive surgical skills<br />
simulation. Endoscopy. 1995;27:81-85.<br />
51. Derossis AM, Fried GM, Abrahamowicz M, Sigman HH, Barkun JS, Meakins JL.<br />
Development <strong>of</strong> a model for training <strong>and</strong> evaluation <strong>of</strong> laparoscopic skills. Am J Surg.<br />
1998;175:482-487.<br />
52. Keyser EJ, Derossis AM, Antoniuk M, Sigman HH, Fried GM. A simplified simulator for<br />
the training <strong>and</strong> evaluation <strong>of</strong> laparoscopic skills. Surg Endosc. 2000;14:149-153.<br />
53. Scott DJ, Bergen PC, Rege RV, Laycock R, Tesfay ST, Valentine RJ, et al. Laparoscopic<br />
training on bench models: better <strong>and</strong> more cost effective than operating room experience? J<br />
Am Coll Surg. 2000;191:272-283.<br />
54. Rosser JC, Herman B, Risucci DA, Murayama M, Rosser LE, Merrell RC. Effectiveness <strong>of</strong><br />
a CD-ROM multimedia tutorial in transferring cognitive knowledge essential for<br />
laparoscopic skill training. Am J Surg. 2000;179:320-324.<br />
55. Gallagher AG, McClure N, McGuigan J, Crothers I, Browning J. Virtual reality training in<br />
laparoscopic surgery: a preliminary assessment <strong>of</strong> minimally invasive surgical trainer<br />
virtual reality (MIST VR). Endoscopy. 1999;31:310-313.<br />
56. Guidelines for the clinical application <strong>of</strong> laparoscopic biliary tract surgery. Society <strong>of</strong><br />
American Gastrointestinal Endoscopic Surgeons. Surg Endosc. 2000;14:771-772.<br />
57. Guidelines for granting <strong>of</strong> privileges for laparoscopic <strong>and</strong>/or thoracoscopic general<br />
surgery. Society <strong>of</strong> American Gastrointestinal Endoscopic Surgeons (SAGES). Surg<br />
Endosc. 1998;12:379-380.<br />
58. Guidelines for surgical treatment <strong>of</strong> gastroesophageal reflux disease (GERD). Society <strong>of</strong><br />
American Gastrointestinal Endoscopic Surgeons (SAGES). Surg Endosc. 1998;12:186-188.<br />
219
59. Fowler DL, Hogle N. The impact <strong>of</strong> a full-time director <strong>of</strong> minimally invasive surgery:<br />
clinical practice, education, <strong>and</strong> research. Surg Endosc. 2000;14:444-447.<br />
60. Sequeira R, Weinbaum F, Satterfield J, Chassin J, Mock L. Credentialing physicians for<br />
new technology: the physician's learning curve must not harm the patient. Am Surg.<br />
1994;60:821-823.<br />
61. Deaton DH, Balch D, Kesler C, Bogey WM, Powell CS. Telemedicine <strong>and</strong> endovascular<br />
aortic grafting. Am J Surg. 1999;177:75-77.<br />
62. Moore RG, Adams JB, Partin AW, Docimo SG, Kavoussi LR. Telementoring <strong>of</strong><br />
laparoscopic procedures: initial clinical experience. Surg Endosc. 1996;10:107-110.<br />
63. Guidelines for submission <strong>of</strong> continuing medical education seeking SAGES endorsement<br />
for courses in laparoscopic surgery. Society <strong>of</strong> American Gastrointestinal Endoscopic<br />
Surgeons (SAGES). Surg Endosc. 1993;7:372-373.<br />
64. Guidelines for diagnostic laparoscopy.SAGES guidelines. Society <strong>of</strong> American<br />
Gastrointestinal Endoscopic Surgeons. Surg Endosc. 1999;13:202-203.<br />
65. Guidelines for laparoscopic surgery during pregnancy. Surg Endosc. 1998;12:189-190.<br />
66. Guidelines for <strong>of</strong>fice endoscopic services. Society <strong>of</strong> American Gastrointestinal<br />
Endoscopic Surgeons (SAGES). Surg Endosc. 1998;12:191-192.<br />
220
Chapter 20. Prevention <strong>of</strong> Surgical Site Infections<br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Subchapter 20.1. Prophylactic Antibiotics<br />
Background<br />
Surgical site infections (SSI) include superficial incisional infections, infections <strong>of</strong> the<br />
deep incision space <strong>and</strong> organ space infections. 1,2 A large body <strong>of</strong> evidence supports the premise<br />
that SSIs can be prevented through administration <strong>of</strong> appropriate prophylactic antibiotics. Two<br />
national organizations, the Federal Centers for Disease Control <strong>and</strong> Prevention (CDC) <strong>and</strong> the<br />
American Society for Health System Pharmacists (ASHP), have recently synthesized this vast<br />
literature to produce comprehensive guidelines regarding the administration <strong>of</strong> prophylactic<br />
antibiotics across a broad range <strong>of</strong> procedures. 3,4 Because <strong>of</strong> the breadth <strong>of</strong> this literature, we<br />
limited the focus <strong>of</strong> this review <strong>of</strong> strategies to prevent SSIs to adult surgery <strong>and</strong> procedures that<br />
typically occur in the operating room (as opposed to procedures such as endoscopy,<br />
interventional cardiology, or radiology procedures).<br />
Practice Description<br />
Antimicrobial prophylaxis refers to a brief course <strong>of</strong> an antimicrobial agent administered<br />
just before an operation begins in order to reduce intraoperative microbial contamination to a<br />
level that will not overwhelm host defenses <strong>and</strong> result in infection. 4 To maximize the benefits <strong>of</strong><br />
antimicrobial prophylactic, the agent used should be safe, inexpensive, <strong>and</strong> bactericidal with an<br />
in vitro spectrum that covers the most probable contaminants for the operation. 4 Administration,<br />
usually by intravenous infusion, should be timed so that a bactericidal concentration is present in<br />
serum <strong>and</strong> tissues by the time the skin is incised. 5 This practice is now st<strong>and</strong>ard <strong>of</strong> care <strong>and</strong><br />
recommended by pr<strong>of</strong>essional societies. 6 Therapeutic levels in serum <strong>and</strong> tissues should be<br />
maintained until, at most, a few hours after the incision is closed in the operating room. 4<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Surgical site infections are a common complication <strong>of</strong> care, occurring in 2-5% <strong>of</strong> patients<br />
after clean extra-abdominal surgeries (eg, thoracic <strong>and</strong> orthopedic surgery) <strong>and</strong> in up to 20% <strong>of</strong><br />
patients undergoing intra-abdominal procedures. 7-12 Studies following patients into the postdischarge<br />
period have reported even higher rates <strong>of</strong> postoperative infection. 13-16 These<br />
complications increase morbidity for patients <strong>and</strong> consume substantial additional resources. 17-21<br />
Opportunities for Impact<br />
Approximately 80-90% <strong>of</strong> surgical patients receive some kind <strong>of</strong> antibiotic prophylaxis,<br />
though recent studies have shown that choice <strong>of</strong> regimen, timing <strong>of</strong> administration or duration <strong>of</strong><br />
prophylaxis is inappropriate in approximately 25-50% <strong>of</strong> cases. 22-27<br />
221
Study Designs <strong>and</strong> Outcomes<br />
As previously noted, the literature on prophylactic antibiotics is extensive. Therefore, the<br />
review was limited to evidence from Level 1A study designs. We identified 9 relevant studies<br />
examining the use <strong>of</strong> prophylactic antibiotics to prevent surgical site infections: 7 meta-analyses<br />
<strong>and</strong> 2 systematic reviews. 28-36 (Tables 20.1.1 <strong>and</strong> 20.1.2) These reviews were <strong>of</strong> high quality <strong>and</strong><br />
limited their source material to r<strong>and</strong>omized controlled trials. Although additional r<strong>and</strong>omized<br />
trials have been published since these reviews were performed, updating the results <strong>of</strong> each<br />
review was beyond the scope <strong>of</strong> this project. All studies examined measured rates <strong>of</strong> site<br />
infection directly (Level 1), using previously published definitions to allow comparability. In<br />
addition, the rates <strong>of</strong> sepsis, length <strong>of</strong> stay, <strong>and</strong> physiologic measures were reported. One metaanalysis<br />
31 <strong>and</strong> one systematic review 33 combined rates <strong>of</strong> several relevant infectious outcomes.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
All studies showed a marked reduction in the odds or relative risk <strong>of</strong> SSI when antibiotic<br />
prophylaxis was employed. None <strong>of</strong> the meta-analyses reviewed explicitly examined the timing<br />
<strong>of</strong> prophylaxis, although many studies pooled data from investigations <strong>of</strong> antibiotic regimens<br />
administered in the immediate preoperative period, (ie, within minutes to an hour <strong>of</strong> initial<br />
incision). Two meta-analyses in our review 29,31 suggested a trend towards lower rates <strong>of</strong><br />
infection with use <strong>of</strong> broader-spectrum antibiotic prophylaxis, such as third generation<br />
cephalosporins. When compared with single dose prophylaxis, multiple dose prophylaxis<br />
generally did not result in significant additional benefit. 29,30,35 In fact, Tanos et al found the odds<br />
<strong>of</strong> SSI were significantly less with single dose prophylaxis. 31 Gillespie et al reported a greater<br />
relative risk <strong>of</strong> infection with single dose prophylaxis with a short-acting antibiotic when<br />
compared with multiple dose prophylaxis. 36 However, the risk <strong>of</strong> infection with single dose<br />
prophylaxis using long-acting antibiotics did not differ significantly from that seen with<br />
multiple-dose regimens.<br />
Potential for Harm<br />
None <strong>of</strong> the meta-analyses analyzed reported rates <strong>of</strong> adverse events (such as allergic<br />
reactions or nosocomial infections) associated with antibiotic prophylaxis <strong>of</strong> any type or<br />
duration. Both <strong>of</strong> the systematic reviews 33,36 noted a trend towards more frequent adverse events<br />
with the use <strong>of</strong> antibiotic prophylaxis. Authors <strong>of</strong> both systematic reviews observed that these<br />
events were reported rarely <strong>and</strong> that variation in the definition <strong>of</strong> “adverse events” across studies<br />
made pooling results difficult.<br />
Infection with Clostridium difficile affects a large number <strong>of</strong> hospitalized patients <strong>and</strong> has<br />
significant clinical <strong>and</strong> economic implications. As many as 16% <strong>of</strong> C. difficile colitis cases in<br />
surgical patients can be attributed to prophylaxis alone, 37 with higher risk for this complication<br />
among patients receiving broad-spectrum antibiotics or prolonged courses <strong>of</strong> therapy. Shortening<br />
the duration <strong>of</strong> antibiotic administration may reduce potential risks <strong>of</strong> prophylaxis (see Chapter<br />
14). Emergence <strong>of</strong> other types <strong>of</strong> resistant pathogens is an additional theoretical concern <strong>of</strong><br />
inappropriate antibiotic prophylaxis; our literature search found no data describing effect <strong>of</strong><br />
antibiotic prophylaxis on population-level incidence <strong>of</strong> these pathogens.<br />
222
Costs <strong>and</strong> Implementation<br />
A number <strong>of</strong> studies have evaluated strategies for improving compliance with<br />
recommended practices for perioperative antibiotic prophylaxis. These include chart audit with<br />
feedback, 38 computerized decision support, 23, 39-42 dissemination <strong>of</strong> guidelines, 43 total quality<br />
management (TQM) <strong>and</strong> continuous quality improvement (CQI) techniques, 44-47 provider<br />
education programs, 48,49 <strong>and</strong> comprehensive efforts by an infection control team. 50 Another<br />
promising <strong>and</strong> easily implemented method is to delegate the administration <strong>of</strong> prophylactic<br />
22, 25, 48<br />
antibiotics to the anesthesia team or the holding room nursing staff.<br />
Costs for systems to increase appropriate use <strong>of</strong> antibiotics will likely be <strong>of</strong>fset by<br />
savings due to prevented infections. However formal analyses <strong>of</strong> the cost-effectiveness <strong>of</strong><br />
specific programs to improve prophylaxis have not been reported.<br />
Comment<br />
For many surgical procedures there is clear evidence supporting the use <strong>of</strong> antibiotic<br />
prophylaxis, administered in a timely manner, to prevent surgical site infections. The reviews<br />
suggest that broader spectrum antibiotics may be superior to limited-spectrum antibiotics for<br />
intra-abdominal or gynecologic surgeries. In addition, single-dose antibiotic prophylaxis appears<br />
to be at least as effective as multiple-dose regimens for a broad range <strong>of</strong> surgical procedures <strong>and</strong><br />
may pose less risk to patients in terms <strong>of</strong> adverse events (eg, C. difficile colitis) <strong>and</strong> less risk to<br />
the population in terms <strong>of</strong> microbial resistance.<br />
Future research will continue to address what prophylactic regimens are most effective<br />
for various surgical procedures. Investigation should also focus on methods to improve<br />
compliance. The optimal strategies for implementation will likely vary from institution to<br />
institution.<br />
223
Table 20.1.1. Meta-analyses examining antibiotic prophylaxis*<br />
Study Trials<br />
Included<br />
Kreter,<br />
1992 35<br />
McDonal<br />
d, 1998 30<br />
Meijer,<br />
1990 29<br />
Mittendor<br />
f, 1993 28<br />
Sharma,<br />
2000 34<br />
Tanos,<br />
1994 31<br />
Surgical Procedures,<br />
Antibiotics<br />
28 Cardiothoracic surgery;<br />
cephalosporins<br />
28 Multiple types <strong>of</strong><br />
surgery;<br />
multiple antibiotics<br />
42 Biliary surgery;<br />
cephalosporins<br />
25 Abdominal<br />
hysterectomy;<br />
multiple antibiotics<br />
6 Percutaneous<br />
gastrostomy;<br />
multiple antibiotics<br />
17 Abdominal<br />
hysterectomy;<br />
cephalosporins<br />
Results: Odds Ratio or Relative Risk <strong>of</strong><br />
Infection (95% CI)<br />
ƒ Cefazolin vs. placebo: OR 0.2 (0.10-0.48).<br />
ƒ Cefazolin vs. cefuroxime or cefam<strong>and</strong>ole:<br />
OR 1.6 (1.03-2.45)<br />
• Single dose vs. multiple dose regimen: no<br />
significant difference<br />
ƒ Single dose vs. multiple dose antibiotics (all<br />
studies): OR 1.06 (0.89-1.25)<br />
ƒ Duration <strong>of</strong> multiple dose regimen 24<br />
hours: OR 1.08 (0.86-1.36)<br />
ƒ Antibiotic vs. placebo: OR 0.30 (0.23-0.38)<br />
ƒ Cephalosporin I vs. cephalosporin II or III:<br />
OR 1.18 (0.69-2)†<br />
• Single dose vs. multiple dose regimen: OR<br />
0.80 (0.4-1.6)<br />
ƒ Antibiotic vs. placebo (all studies): OR 0.35<br />
(0.27-0.5); p
Wilson,<br />
1992 51<br />
21 Multiple types <strong>of</strong><br />
surgery;<br />
multiple antibiotics<br />
225<br />
• Single dose vs. multiple dose regimen: OR<br />
0.37 (0.3-0.5)<br />
ƒ Amoxicillin-clavulanic acid vs. other<br />
antibiotics (all studies): OR 0.84 (0.68-<br />
1.04)<br />
ƒ Trend favoring amoxicillin-clavulanic acid<br />
for biliary <strong>and</strong> gynecologic surgery<br />
* CI indicates confidence interval; NNT, number needed to treat; OR, odds ratio, <strong>and</strong> RR,<br />
relative risk.<br />
† Roman numerals I, II, III indicate generation <strong>of</strong> cephalosporin antibiotics.<br />
‡ P values were reported in article; OR were approximated based on figures.
Table 20.1.2. Systematic reviews <strong>of</strong> antibiotic prophylaxis*<br />
Study Trials<br />
Included<br />
Gillespie,<br />
2000 36<br />
Smaill,<br />
2000 33<br />
Surgical<br />
Procedures;<br />
Antibiotics<br />
Results: Relative Risk <strong>of</strong> Infection (95% CI)<br />
48 Long bone Single dose antibiotic vs. placebo<br />
fractures;<br />
Deep wound infection: RR 0.40 (0.24-0.67)<br />
multiple<br />
antibiotics<br />
Superficial wound infection: RR 0.69 (0.50-0.95)<br />
Urinary tract infection: RR 0.63 (0.53-0.76)<br />
Pneumonia: RR 0.46 (0.33-0.65)<br />
Multiple dose antibiotic vs. placebo:<br />
Deep wound infection: RR 0.36 (0.21-0.65)<br />
Superficial wound infection: RR 0.48 (0.28-0.81)<br />
Urinary tract infection: RR 0.66 (0.4-1.0)<br />
Pneumonia: RR 0.81 (0.41-1.63)<br />
Adverse events: RR 1.83 (0.96-3.50)<br />
Single dose short-acting antibiotic vs. multiple doses same<br />
agent up to 24 hours after surgery<br />
Deep wound infection: RR 7.98 (1.01-62.0)<br />
Superficial wound infection: RR 4.82 (1.08-21.6)<br />
Urinary tract infection: RR 1.81 (1.01-3.23)<br />
Single dose long-acting antibiotic vs. any multiple dose<br />
regimen lasting more than 24 hours<br />
Deep wound infection: RR 1.10 (0.22-5.34)<br />
Superficial wound infection: RR 0.57 (0.17-1.93)<br />
Multiple doses administered over 24 hours or less vs.<br />
longer therapy<br />
Deep wound infection: RR 1.1 (0.22-5.34)<br />
Superficial wound infection: RR 0.57 (0.17-1.93)<br />
Oral vs. parenteral prophylaxis<br />
Insufficient data (single underpowered study)<br />
66 Cesarean section; Impact <strong>of</strong> antibiotic prophylaxis on …<br />
multiple -Combined outcomes <strong>of</strong> fever, wound infection, sepsis <strong>and</strong><br />
antibiotics endometritis:<br />
Elective Cesarean section: RR 0.25 (0.11-0.55)<br />
Emergent Cesarean section: RR 0.39 (0.33-0.46)<br />
Unspecified/nonelective: RR 0.37 (0.32-0.42)<br />
226
All Cesarean section: RR 0.37 (0.33-0.42)<br />
-Maternal side effects: RR 1.96 (0.86-4.49)<br />
-Length <strong>of</strong> stay: 0.34 fewer days in hospital (0.17-0.52)<br />
* CI indicates confidence interval; RR, relative risk.<br />
227
References<br />
1. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guideline for the Prevention<br />
<strong>of</strong> Surgical Site Infection. Hospital Infection Program, Centers for Disease Control <strong>and</strong><br />
Prevention. Available at: http://www.cdc.gov/ncidod/hip/. Accessed April 29, 2001.<br />
2. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guideline for prevention <strong>of</strong><br />
surgical site infection, 1999. Hospital Infection Control <strong>Practices</strong> Advisory Committee.<br />
Infect Control Hosp Epidemiol. 1999;20:250-278.<br />
3. ASHP Therapeutic Guidelines on Antimicrobial Prophylaxis in Surgery. American Society<br />
<strong>of</strong> Health-System Pharmacists. Am J Health Syst Pharm. 1999;56:1839-1888.<br />
4. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guideline for Prevention <strong>of</strong><br />
Surgical Site Infection, 1999. Centers for Disease Control <strong>and</strong> Prevention (CDC) Hospital<br />
Infection Control <strong>Practices</strong> Advisory Committee. Am J Infect Control. 1999;27:97-132.<br />
5. Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL, Burke JP. The timing <strong>of</strong><br />
prophylactic administration <strong>of</strong> antibiotics <strong>and</strong> the risk <strong>of</strong> surgical-wound infection. N Engl<br />
J Med. 1992;326:281-286.<br />
6. Dellinger EP, Gross PA, Barrett TL, et al. Quality st<strong>and</strong>ard for antimicrobial prophylaxis in<br />
surgical procedures. The Infectious Diseases Society <strong>of</strong> America. Infect Control Hosp<br />
Epidemiol. 1994;15:182-188.<br />
7. Delgado-Rodriguez M, Sillero-Arenas M, Medina-Cuadros M, Martinez-Gallego G.<br />
Nosocomial infections in surgical patients: comparison <strong>of</strong> two measures <strong>of</strong> intrinsic patient<br />
risk. Infect Control Hosp Epidemiol. 1997;18:19-23.<br />
8. Horan TC, Culver DH, Gaynes RP, Jarvis WR, Edwards JR, Reid CR. Nosocomial<br />
infections in surgical patients in the United States, January 1986-June 1992. National<br />
Nosocomial Infections Surveillance (NNIS) System. Infect Control Hosp Epidemiol.<br />
1993;14:73-80.<br />
9. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions <strong>of</strong><br />
nosocomial surgical site infections, 1992: a modification <strong>of</strong> CDC definitions <strong>of</strong> surgical<br />
wound infections. Infect Control Hosp Epidemiol. 1992;13:606-608.<br />
10. Horan TC, Emori TG. Definitions <strong>of</strong> key terms used in the NNIS System. Am J Infect<br />
Control. 1997;25:112-116.<br />
11. Wallace WC, Cinat M, Gornick WB, Lekawa ME, Wilson SE. Nosocomial infections in<br />
the surgical intensive care unit: a difference between trauma <strong>and</strong> surgical patients. Am<br />
Surg. 1999;65:987-990.<br />
12. Scheel O, Stormark M. National prevalence survey on hospital infections in Norway. J<br />
Hosp Infect. 1999;41:331-335.<br />
13. S<strong>and</strong>s K, Vineyard G, Platt R. Surgical site infections occurring after hospital discharge. J<br />
Infect Dis. 1996;173:963-970.<br />
14. Waddell TK, Rotstein OD. Antimicrobial prophylaxis in surgery. Committee on<br />
Antimicrobial Agents, Canadian Infectious Disease Society. CMAJ. 1994;151:925-931.<br />
15. Medina-Cuadros M, Sillero-Arenas M, Martinez-Gallego G, Delgado-Rodriguez M.<br />
Surgical wound infections diagnosed after discharge from hospital: epidemiologic<br />
differences with in-hospital infections. Am J Infect Control. 1996;24:421-428.<br />
16. Lynch W, Davey PG, Malek M, Byrne DJ, Napier A. Cost-effectiveness analysis <strong>of</strong> the use<br />
<strong>of</strong> chlorhexidine detergent in preoperative whole-body disinfection in wound infection<br />
prophylaxis. J Hosp Infect. 1992;21:179-191.<br />
228
17. Vegas AA, Jodra VM, Garcia ML. Nosocomial infection in surgery wards: a controlled<br />
study <strong>of</strong> increased duration <strong>of</strong> hospital stays <strong>and</strong> direct cost <strong>of</strong> hospitalization. Eur J<br />
Epidemiol. 1993;9:504-510.<br />
18. Kirkl<strong>and</strong> KB, Briggs JP, Trivette SL, Wilkinson WE, Sexton DJ. The impact <strong>of</strong> surgicalsite<br />
infections in the 1990s: attributable mortality, excess length <strong>of</strong> hospitalization, <strong>and</strong><br />
extra costs. Infect Control Hosp Epidemiol. 1999;20:725-730.<br />
19. Collins TC, Daley J, Henderson WH, Khuri SF. Risk factors for prolonged length <strong>of</strong> stay<br />
after major elective surgery. Ann Surg. 1999;230:251-259.<br />
20. Asensio A, Torres J. Quantifying excess length <strong>of</strong> postoperative stay attributable to<br />
infections: a comparison <strong>of</strong> methods. J Clin Epidemiol. 1999;52:1249-1256.<br />
21. Wong ES. The price <strong>of</strong> a surgical-site infection: more than just excess length <strong>of</strong> stay. Infect<br />
Control Hosp Epidemiol. 1999;20:722-724.<br />
22. Silver A, Eichorn A, Kral J, et al. Timeliness <strong>and</strong> use <strong>of</strong> antibiotic prophylaxis in selected<br />
inpatient surgical procedures. The Antibiotic Prophylaxis Study Group. Am J Surg.<br />
1996;171:548-552.<br />
23. Larsen RA, Evans RS, Burke JP, Pestotnik SL, Gardner RM, Classen DC. Improved<br />
perioperative antibiotic use <strong>and</strong> reduced surgical wound infections through use <strong>of</strong> computer<br />
decision analysis. Infect Control Hosp Epidemiol. 1989;10:316-320.<br />
24. Finkelstein R, Reinhertz G, Embom A. Surveillance <strong>of</strong> the use <strong>of</strong> antibiotic prophylaxis in<br />
surgery. Isr J Med Sci. 1996;32:1093-1097.<br />
25. Matuschka PR, Cheadle WG, Burke JD, Garrison RN. A new st<strong>and</strong>ard <strong>of</strong> care:<br />
administration <strong>of</strong> preoperative antibiotics in the operating room. Am Surg. 1997;63:500-<br />
503.<br />
26. Gorecki P, Schein M, Rucinski JC, Wise L. Antibiotic administration in patients<br />
undergoing common surgical procedures in a community teaching hospital: the chaos<br />
continues. World J Surg. 1999;23:429-432.<br />
27. Zoutman D, Chau L, Watterson J, Mackenzie T, Djurfeldt M. A Canadian survey <strong>of</strong><br />
prophylactic antibiotic use among hip-fracture patients. Infect Control Hosp Epidemiol.<br />
1999;20:752-755.<br />
28. Mittendorf R, Aronson MP, Berry RE, et al. Avoiding serious infections associated with<br />
abdominal hysterectomy: a meta-analysis <strong>of</strong> antibiotic prophylaxis. Am J Obstet Gynecol.<br />
1993;169:1119-1124.<br />
29. Meijer WS, Schmitz PI, Jeekel J. Meta-analysis <strong>of</strong> r<strong>and</strong>omized, controlled clinical trials <strong>of</strong><br />
antibiotic prophylaxis in biliary tract surgery. Br J Surg. 1990;77:283-290.<br />
30. McDonald M, Grabsch E, Marshall C, Forbes A. Single- versus multiple-dose<br />
antimicrobial prophylaxis for major surgery: a systematic review. Aust N Z J Surg.<br />
1998;68:388-396.<br />
31. Tanos V, Rojansky N. Prophylactic antibiotics in abdominal hysterectomy. J Am Coll Surg.<br />
1994;179:593-600.<br />
32. Song F, Glenny AM. Antimicrobial prophylaxis in colorectal surgery: a systematic review<br />
<strong>of</strong> r<strong>and</strong>omized controlled trials. Br J Surg. 1998;85:1232-1241. Published erratum appears<br />
in Br J Surg 1999;1286:1280.<br />
33. Smaill F, H<strong>of</strong>meyr GJ. Antibiotic prophylaxis for cesarean section. In: The Cochrane<br />
Library, Issue 2, 2000. Oxford: Update S<strong>of</strong>tware.<br />
34. Sharma VK, Howden CW. Meta-analysis <strong>of</strong> r<strong>and</strong>omized, controlled trials <strong>of</strong> antibiotic<br />
prophylaxis before percutaneous endoscopic gastrostomy. Am J Gastroenterol.<br />
2000;95:3133-3136.<br />
229
35. Kreter B, Woods M. Antibiotic prophylaxis for cardiothoracic operations. Meta-analysis <strong>of</strong><br />
thirty years <strong>of</strong> clinical trials. J Thorac Cardiovasc Surg. 1992;104:590-599.<br />
36. Gillespie WJ, Walenkamp G. Antibiotic prophylaxis for surgery for proximal femoral <strong>and</strong><br />
other closed long bone fractures. In: The Cochrane Library, Issue 2, 2000. Oxford: Update<br />
S<strong>of</strong>tware; 2000.<br />
37. Crabtree TD, Pelletier SJ, Gleason TG, Pruett TL, Sawyer RG. Clinical characteristics <strong>and</strong><br />
antibiotic utilization in surgical patients with Clostridium difficile-associated diarrhea. Am<br />
Surg. 1999;65:507-511.<br />
38. Zhanel GG, Gin AS, Przybylo A, Louie TJ, Otten NH. Effect <strong>of</strong> interventions on<br />
prescribing <strong>of</strong> antimicrobials for prophylaxis in obstetric <strong>and</strong> gynecologic surgery. Am J<br />
Hosp Pharm. 1989;46:2493-2496.<br />
39. Pestotnik SL, Evans RS, Burke JP, Gardner RM, Classen DC. Therapeutic antibiotic<br />
monitoring: surveillance using a computerized expert system. Am J Med. 1990;88:43-48.<br />
40. Pestotnik SL, Classen DC, Evans RS, Burke JP. Implementing antibiotic practice<br />
guidelines through computer-assisted decision support: clinical <strong>and</strong> financial outcomes.<br />
Ann Intern Med. 1996;124:884-890.<br />
41. Evans RS, Pestotnik SL, Classen DC, et al. A computer-assisted management program for<br />
antibiotics <strong>and</strong> other antiinfective agents. N Engl J Med. 1998;338:232-238.<br />
42. Evans RS, Pestotnik SL, Burke JP, Gardner RM, Larsen RA, Classen DC. Reducing the<br />
duration <strong>of</strong> prophylactic antibiotic use through computer monitoring <strong>of</strong> surgical patients.<br />
DICP. 1990;24:351-354.<br />
43. Dobrzanski S, Lawley DI, McDermott I, Selby M, Ausobsky JR. The impact <strong>of</strong> guidelines<br />
on peri-operative antibiotic administration. J Clin Pharm Ther. 1991;16:19-24.<br />
44. Koska MT. Using CQI methods to lower postsurgical wound infection rate. Hospitals.<br />
1992;66:62,64.<br />
45. Kroll DA, Brummitt CF, Berry BB. A users group approach to quality improvement across<br />
an integrated healthcare delivery system. J Healthc Qual. 2000;22:39-43.<br />
46. Welch L, Teague AC, Knight BA, Kenney A, Hern<strong>and</strong>ez JE. A quality management<br />
approach to optimizing delivery <strong>and</strong> administration <strong>of</strong> preoperative antibiotics. Clin<br />
Perform Qual Health Care. 1998;6:168-171.<br />
47. Woster PS, Ryan ML, Ginsberg-Evans L, Olson J. Use <strong>of</strong> total quality management<br />
techniques to improve compliance with a medication use indicator. Top Hosp Pharm<br />
Manage. 1995;14:68-77.<br />
48. Gyssens IC, Knape JT, Van Hal G, ver der Meer JW. The anaesthetist as determinant factor<br />
<strong>of</strong> quality <strong>of</strong> surgical antimicrobial prophylaxis. A survey in a university hospital. Pharm<br />
World Sci. 1997;19:89-92.<br />
49. Everitt DE, Soumerai SB, Avorn J, Klapholz H, Wessels M. Changing surgical<br />
antimicrobial prophylaxis practices through education targeted at senior department<br />
leaders. Infect Control Hosp Epidemiol. 1990;11:578-583.<br />
50. McConkey SJ, L'Ecuyer PB, Murphy DM, Leet TL, Sundt TM, Fraser VJ. Results <strong>of</strong> a<br />
comprehensive infection control program for reducing surgical-site infections in coronary<br />
artery bypass surgery. Infect Control Hosp Epidemiol. 1999;20:533-538.<br />
51. Wilson AP, Shrimpton S, Jaderberg M. A meta-analysis <strong>of</strong> the use <strong>of</strong> amoxycillinclavulanic<br />
acid in surgical prophylaxis. J Hosp Infect. 1992;22:9-21.<br />
230
Subchapter 20.2. Perioperative Normothermia<br />
Background<br />
The body temperature <strong>of</strong> patients may fall by 1 to 1.5°C during the first hour <strong>of</strong> general<br />
anesthesia. 1 Regional anesthesia also typically causes core hypothermia. 2 Intraoperative<br />
hypothermia impairs immune function (especially oxidative killing by neutrophils) <strong>and</strong> results in<br />
dermal vasoconstriction <strong>and</strong> reduced blood flow to surgical sites, which further increases the risk<br />
<strong>of</strong> surgical site infection by lowering tissue oxygen tension. 3 Hypothermia also results in reduced<br />
platelet function, shivering associated with patient discomfort <strong>and</strong> activation <strong>of</strong> the sympathetic<br />
nervous system, <strong>and</strong> adverse cardiac events. 2<br />
Practice Description<br />
Normal core temperature can be maintained during surgery through use <strong>of</strong> active<br />
measures including warmed intravenous fluids <strong>and</strong> inspired gases, as well as forced air warming.<br />
The latter involves an air blanket placed over the patient that circulates air warmed to 40°C.<br />
Water blankets may also be used, but are not as effective in maintaining body temperature. 4<br />
<strong>Patient</strong> temperature is monitored using conventional thermometer probes, with active measures<br />
adjusted to maintain core temperature near 36.5°C. Any method or combination <strong>of</strong> methods that<br />
maintains the target core temperature appears to have the same effect. 2<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
See Subchapter 20.1.<br />
Opportunities for impact<br />
Attention to patient temperature is st<strong>and</strong>ard <strong>of</strong> care in intraoperative anesthesia<br />
management. * However, there are no data on the extent to which active warming measures are<br />
currently used perioperatively.<br />
* The American Society <strong>of</strong> Anesthesiologists’ St<strong>and</strong>ards for Basic Anesthesia Monitoring notes<br />
“Every patient receiving anesthesia shall have temperature monitored when clinically significant<br />
changes in body temperature are intended, anticipated or suspected.” 5<br />
231
Study Designs <strong>and</strong> Outcomes<br />
We identified one r<strong>and</strong>omized controlled trial 3 <strong>and</strong> one retrospective cohort study 6<br />
evaluating the effect <strong>of</strong> active warming interventions on the rate <strong>of</strong> wound infection (Level 1<br />
outcome). (Table 20.2.1). Wound infection was either defined as “suppuration requiring removal<br />
<strong>of</strong> sutures” 3 or as in previously published definitions. 7<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Kurz et al performed a r<strong>and</strong>omized controlled trial <strong>of</strong> active warming in the<br />
intraoperative care <strong>of</strong> patients undergoing elective colectomy. All patients received aggressive<br />
perioperative hydration <strong>and</strong> intravenous opioids for pain relief, in an effort to maximize wound<br />
perfusion. <strong>Patient</strong>s in the normothermia arm experienced a 68% reduction in the rate <strong>of</strong> wound<br />
infection, lower wound infection scores (as defined by the elements <strong>of</strong> the acronym ASEPSIS:<br />
Additional treatment, Serous discharge, Erythema, Purulent exudate, Separation <strong>of</strong> deep tissues,<br />
Isolation <strong>of</strong> bacteria, <strong>and</strong> duration <strong>of</strong> inpatient Stay), <strong>and</strong> shorter length <strong>of</strong> hospitalization. 3<br />
While the relatively high infection rate (19% <strong>of</strong> control group in this university-based population<br />
with a substantial degree <strong>of</strong> underlying disease) <strong>and</strong> suboptimal antibiotic prophylaxis<br />
(antibiotics continued for about 4 days postoperatively; see Subchapter 20.1) do not invalidate<br />
the study results, they do limit their generalizability.<br />
In a retrospective cohort study based on chart reviews <strong>of</strong> 150 patients undergoing elective<br />
colectomy, Barone et al noted no independent association between intraoperative hypothermia<br />
(defined as temperature less than
Comment<br />
Given the evidence <strong>of</strong> effectiveness, the low potential for harm, <strong>and</strong> the simplicity <strong>of</strong> the<br />
intervention (including the ready availability <strong>of</strong> the equipment), maintenance <strong>of</strong> perioperative<br />
normothermia seems a promising practice to improve patient safety. The methodologically<br />
stronger <strong>of</strong> the 2 studies reviewed showed clear benefits. However, some <strong>of</strong> its benefits may not<br />
be generalizable to patient populations undergoing other procedures. For example, intraoperative<br />
hypothermia may have little impact on wound infections in patients undergoing cesarean<br />
section. 14 Thus, additional study <strong>of</strong> the practice is needed in other settings. Furthermore, for<br />
some procedures hypothermia is likely to protect patients. Core temperature is <strong>of</strong>ten intentionally<br />
reduced to protect the myocardium <strong>and</strong> central nervous system during certain cardiac <strong>and</strong><br />
neurosurgical procedures. 2,12,15 In such cases the potential benefits <strong>of</strong> normothermia may not<br />
outweigh the associated risks.<br />
Table 20.2.1. Summary <strong>of</strong> studies reporting effectiveness <strong>of</strong> perioperative normothermia*<br />
Study Study Population; Intervention Study Design,<br />
Outcomes<br />
Kurz,<br />
1996 3<br />
Barone,<br />
1999 6<br />
200 patients (104 normothermia, 96<br />
hypothermia) undergoing, elective<br />
colectomy in multicenter study;<br />
warmed gases, fluids <strong>and</strong> forced arm<br />
during operation vs. usual care<br />
150 patients (101 normothermia, 49<br />
hypothermia) undergoing elective<br />
colectomy at a single community<br />
hospital; no formal intervention<br />
(retrospective chart review,<br />
warming devices were used in 90%<br />
<strong>of</strong> patients)<br />
Level 1,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
233<br />
Results<br />
Wound infection rate: 6% vs. 19%<br />
(p=0.009)<br />
ASEPSIS score: 7 vs. 13 (p=0.002)<br />
Days to sutures out: 9.9 vs. 10.9<br />
(p=0.002)<br />
Taking nutrition orally: 5.6 vs. 6.5 days<br />
(p=0.006)<br />
Length <strong>of</strong> stay: 12 vs. 15 days (p=0.001)<br />
Wound infection rate: 12% in both<br />
groups<br />
Multivariate models: no significant<br />
association between hypothermia<br />
<strong>and</strong> wound infection or length <strong>of</strong><br />
stay<br />
* ASEPSIS indicates Additional treatment, Serous discharge, Erythema, Purulent exudate, Separation <strong>of</strong><br />
deep tissues, Isolation <strong>of</strong> bacteria, <strong>and</strong> duration <strong>of</strong> inpatient Stay. 7<br />
References<br />
1. Matsukawa T, Sessler DI, Sessler AM, et al. Heat flow <strong>and</strong> distribution during induction <strong>of</strong><br />
general anesthesia. Anesthesiology. 1995;82:662-673.<br />
2. Sessler DI. Mild perioperative hypothermia. N Engl J Med. 1997;336:1730-1737.<br />
3. Kurz A, Sessler DI, Lenhardt R. Perioperative normothermia to reduce the incidence <strong>of</strong><br />
surgical-wound infection <strong>and</strong> shorten hospitalization. Study <strong>of</strong> Wound Infection <strong>and</strong><br />
Temperature Group. N Engl J Med. 1996;334:1209-1215.
4. Kurz A, Kurz M, Poeschl G, Faryniak B, Redl G, Hackl W. Forced-air warming maintains<br />
intraoperative normothermia better than circulating-water mattresses. Anesth Analg.<br />
1993;77:89-95.<br />
5. American Society <strong>of</strong> Anesthesiologists. St<strong>and</strong>ards <strong>of</strong> the American Society <strong>of</strong><br />
Anesthesiologists. Available at: http://www.asahq.org/St<strong>and</strong>ards/02.html. Accessed June 6,<br />
2001.<br />
6. Barone JE, Tucker JB, Cecere J, et al. Hypothermia does not result in more complications<br />
after colon surgery. Am Surg. 1999;65:356-359.<br />
7. Wilson AP, Treasure T, Sturridge MF, Gruneberg RN. A scoring method (ASEPSIS) for<br />
postoperative wound infections for use in clinical trials <strong>of</strong> antibiotic prophylaxis. Lancet.<br />
1986;1:311-313.<br />
8. Sessler DI, Kurz A, Lenhardt R. Re: Hypothermia reduces resistance to surgical wound<br />
infections. Am Surg. 1999;65:1193-1196.<br />
9. Frank SM, Fleisher LA, Breslow MJ, et al. Perioperative maintenance <strong>of</strong> normothermia<br />
reduces the incidence <strong>of</strong> morbid cardiac events. A r<strong>and</strong>omized clinical trial. JAMA.<br />
1997;277:1127-1134.<br />
10. Schmied H, Kurz A, Sessler DI, Kozek S, Reiter A. Mild hypothermia increases blood loss<br />
<strong>and</strong> transfusion requirements during total hip arthroplasty. Lancet. 1996;347:289-292.<br />
11. Winkler M, Akca O, Birkenberg B, et al. Aggressive warming reduces blood loss during<br />
hip arthroplasty. Anesth Analg. 2000;91:978-984.<br />
12. Leslie K, Sessler DI. The implications <strong>of</strong> hypothermia for early tracheal extubation<br />
following cardiac surgery. J Cardiothorac Vasc Anesth. 1998;12:30-34; discussion 41-34.<br />
13. Sigg DC, Houlton AJ, Iaizzo PA. The potential for increased risk <strong>of</strong> infection due to the<br />
reuse <strong>of</strong> convective air-warming/cooling coverlets. Acta Anaesthesiol Sc<strong>and</strong>. 1999;43:173-<br />
176.<br />
14. Munn MB, Rouse DJ, Owen J. Intraoperative hypothermia <strong>and</strong> post-cesarean wound<br />
infection. Obstet Gynecol. 1998;91:582-584.<br />
15. Winfree CH, Baker KZ, Connollly ES. Perioperative normothermia <strong>and</strong> surgical-wound<br />
infection. N Engl J Med. 1996;335:749.<br />
Subchapter 20.3. Supplemental Perioperative Oxygen<br />
Background<br />
Low oxygen content in devitalized tissues predisposes them to bacterial colonization,<br />
which is thought to be a key pathophysiologic step in the initiation <strong>of</strong> surgical site infections. 1<br />
Administration <strong>of</strong> high concentrations <strong>of</strong> oxygen increases wound oxygen tension, allowing for<br />
more effective neutrophil function <strong>and</strong> the potential for reduced infection rates. 2<br />
Practice Description<br />
The practice <strong>of</strong> perioperative oxygen supplementation involves administration <strong>of</strong> 80%<br />
oxygen <strong>and</strong> 20% nitrogen by endotracheal tube intraoperatively <strong>and</strong> by sealed mask <strong>and</strong><br />
manifold system or conventional non-rebreather mask for the first two hours <strong>of</strong> recovery.<br />
Oxygen is increased to 100% immediately before extubation, with the concentration returned to<br />
80% as soon as deemed safe by the anesthesiologist. 3<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
See Subchapter 20.1.<br />
234
Opportunities for Impact<br />
Administration <strong>of</strong> oxygen is a routine part <strong>of</strong> perioperative care. However the frequency<br />
with which high oxygen concentrations (as described above) are administered is not known.<br />
Study Designs <strong>and</strong> Outcomes<br />
We identified one r<strong>and</strong>omized controlled trial evaluating the effect <strong>of</strong> high concentration<br />
oxygen supplementation on surgical site infections (Table 20.3.1). 3 The primary outcome was<br />
incidence <strong>of</strong> wound infection within 15 days after surgery (Level 1). Wounds were considered<br />
infected when bacteria were cultured from pus expressed from the incision or aspirated from a<br />
loculated collection within the wound. 3<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The clinical characteristics <strong>of</strong> the intervention <strong>and</strong> control groups were similar at<br />
baseline, including risk <strong>of</strong> infection as assessed by a modified Study on the Efficacy <strong>of</strong><br />
Nosocomial Infection Control (SENIC) score (p=0.8) <strong>and</strong> National Nosocomial Infection<br />
Surveillance System (NNISS) score (p=0.86). The incidence <strong>of</strong> wound infection was<br />
significantly less in the intervention group (13/250, 5%) than in the control group (28/250, 11%,<br />
p=0.014). The results remain statistically significant when the study definition <strong>of</strong> “infection” is<br />
broadened to include wounds with pus but no bacterial growth on culture (7% vs. 14%,<br />
p=0.012). Perioperative administration <strong>of</strong> high levels <strong>of</strong> oxygen was associated with a 54%<br />
relative risk reduction (95% CI: 12%-75%) <strong>of</strong> wound infection within 15 days <strong>of</strong> surgery.<br />
ASEPSIS (Additional treatment, Serous discharge, Erythema, Purulent exudate, Separation <strong>of</strong><br />
deep tissues, Isolation <strong>of</strong> bacteria, <strong>and</strong> duration <strong>of</strong> inpatient Stay 4 ) scores were also significantly<br />
better with high levels <strong>of</strong> oxygen (3 vs. 5, p=0.01). Although longer follow-up might have<br />
identified additional wound infections, the authors argue that it was unlikely that these events<br />
would take place preferentially in one group as the proposed therapeutic effect <strong>of</strong> oxygen<br />
appears limited to the immediate perioperative period. 3 Admission to the intensive care unit <strong>and</strong><br />
death were less frequent in the intervention group, but the difference failed to achieve statistical<br />
significance.<br />
Two additional r<strong>and</strong>omized controlled trials <strong>of</strong> perioperative supplemental oxygen were<br />
identified. 5,6 Both found a significant reduction in postoperative nausea <strong>and</strong> vomiting, but neither<br />
study evaluated the effect on wound infections.<br />
Potential for Harm<br />
The study by Greif et al reported no adverse affects related to the intervention. Several<br />
potential risks <strong>of</strong> high oxygen concentrations should be noted. High oxygen concentrations may<br />
present a fire hazard when heated surgical instruments (eg, lasers) are introduced into the<br />
airway. 7-11 Such concentrations can also induce lung injury in certain vulnerable patients 12 or<br />
precipitate atelectasis in patients at risk. 3,13,14 Hyperoxic mixtures may increase oxidative<br />
myocardial injury in patients undergoing cardiopulmonary bypass. 15 Finally, patients who<br />
undergo resuscitation with 100% oxygen may have worsened neurologic outcomes, possibly also<br />
as a result <strong>of</strong> increased oxygen free-radical generation. 16,17<br />
Costs <strong>and</strong> Implementation<br />
The incremental direct costs associated with administering high oxygen concentrations<br />
are minimal, as oxygen delivery systems are elements <strong>of</strong> routine perioperative care <strong>and</strong> employ<br />
equipment readily available in operating rooms.<br />
235
Comment<br />
Administration <strong>of</strong> perioperative oxygen in high concentrations seems a promising<br />
adjunctive therapy: the practice is simple, the equipment needed is readily available, <strong>and</strong> a<br />
multicenter r<strong>and</strong>omized trial has demonstrated its efficacy.<br />
However, there are significant questions about the generalizability <strong>of</strong> the approach to<br />
exp<strong>and</strong>ed populations <strong>of</strong> surgical patients. All patients in the Grief et al study had core<br />
temperature maintained at 36ºC, were aggressively hydrated, <strong>and</strong> had postoperative pain treated<br />
with opioids in order to maximize wound perfusion. To what degree the effectiveness <strong>of</strong> the<br />
practice is affected by changes in these “co-interventions” has not been assessed. There is reason<br />
for concern regarding use <strong>of</strong> high concentrations <strong>of</strong> oxygen in patients undergoing procedures<br />
associated with low blood flow (eg, cardiopulmonary bypass), or in whom local production <strong>of</strong><br />
oxygen free radicals may cause further organ injury (eg, patients with head trauma).<br />
Additionally, questions remain regarding whether modifications to the protocol used<br />
would impart similar or greater benefit. For example, would oxygen administration by nasal<br />
cannula at 10 LPM be as effective as oxygen delivered by a sealed mask? Would longer duration<br />
<strong>of</strong> therapy impart additional benefit? These questions should be answered in future trials.<br />
Table 20.3.1. R<strong>and</strong>omized controlled trial <strong>of</strong> supplemental perioperative oxygen*<br />
Study Study Population Intervention Results†<br />
Greif,<br />
2000 3<br />
500 patients<br />
undergoing colorectal<br />
resection; multicenter<br />
study, 1996-98<br />
80% oxygen, 20%<br />
nitrogen during surgery<br />
<strong>and</strong> the first 2 hours <strong>of</strong><br />
recovery<br />
236<br />
Wound infection:<br />
ARR 0.06 (95% CI, 0.018-0.102)<br />
RR 0.46 (95% CI, 0.25-0.88)<br />
ASEPSIS § score: 3 vs. 5 (p=0.01)<br />
ICU admission: 2.0% vs. 4.8%<br />
(p=0.14)<br />
Mortality: 0.4% vs. 2.4% (p=0.13)<br />
* ARR indicates absolute risk reduction; CI, confidence interval; ICU, intensive care unit; <strong>and</strong><br />
RR, relative risk. The ASEPSIS scoring system incorporates Additional treatment, Serous<br />
discharge, Erythema, Purulent exudate, Separation <strong>of</strong> deep tissues, Isolation <strong>of</strong> bacteria, <strong>and</strong><br />
duration <strong>of</strong> inpatient Stay 4 .<br />
† Outcomes within 15 days <strong>of</strong> surgery, expressed as rates in intervention vs. control groups.<br />
References<br />
1. Hopf HW, Hunt TK, West JM, et al. Wound tissue oxygen tension predicts the risk <strong>of</strong><br />
wound infection in surgical patients. Arch Surg. 1997;132:997-1004.<br />
2. Hopf H, Sessler DI. Routine postoperative oxygen supplementation. Anesth Analg.<br />
1994;79:615-616.<br />
3. Greif R, Akca O, Horn EP, Kurz A, Sessler DI. Supplemental perioperative oxygen to<br />
reduce the incidence <strong>of</strong> surgical- wound infection. Outcomes Research Group. N Engl J<br />
Med. 2000;342:161-167.<br />
4. Wilson AP, Treasure T, Sturridge MF, Gruneberg RN. A scoring method (ASEPSIS) for<br />
postoperative wound infections for use in clinical trials <strong>of</strong> antibiotic prophylaxis. Lancet.<br />
1986;1:311-313.
5. Greif R, Laciny S, Rapf B, Hickle RS, Sessler DI. Supplemental oxygen reduces the<br />
incidence <strong>of</strong> postoperative nausea <strong>and</strong> vomiting. Anesthesiology 1999;91:1246-1252.<br />
6. Goll V, Akca O, Greif R, et al. Ondansetron is no more effective than supplemental<br />
intraoperative oxygen for prevention <strong>of</strong> postoperative nausea <strong>and</strong> vomiting. Anesth Analg.<br />
2001;92:112-117.<br />
7. Healy GB. Complications <strong>of</strong> laser surgery. Otolaryngol Clin North Am. 1983;16:815-820.<br />
8. Hunsaker DH. Anesthesia for microlaryngeal surgery: the case for subglottic jet<br />
ventilation. Laryngoscope. 1994;104:1-30.<br />
9. Aly A, McIlwain M, Duncavage JA. Electrosurgery-induced endotracheal tube ignition<br />
during tracheotomy. Ann Otol Rhinol Laryngol. 1991;100:31-33.<br />
10. Barnes AM, Frantz RA. Do oxygen-enriched atmospheres exist beneath surgical drapes<br />
<strong>and</strong> contribute to fire hazard potential in the operating room? Aana J. 2000;68:153-161.<br />
11. de Richemond AL, Bruley ME. Use <strong>of</strong> supplemental oxygen during surgery is not risk free.<br />
Anesthesiology. 2000;93:583-584.<br />
12. Donat SM, Levy DA. Bleomycin associated pulmonary toxicity: is perioperative oxygen<br />
restriction necessary? J Urol. 1998;160:1347-1352.<br />
13. Akca O, Podolsky A, Eisenhuber E, et al. Comparable postoperative pulmonary atelectasis<br />
in patients given 30% or 80% oxygen during <strong>and</strong> 2 hours after colon resection.<br />
Anesthesiology. 1999;91:991-998.<br />
14. Lentschener C. Prevention <strong>of</strong> atelectasis during general anaesthesia. Lancet. 1995;346:514-<br />
515.<br />
15. Ihnken K, Winkler A, Schlensak C, et al. Normoxic cardiopulmonary bypass reduces<br />
oxidative myocardial damage <strong>and</strong> nitric oxide during cardiac operations in the adult. J<br />
Thorac Cardiovasc Surg. 1998;116:327-334.<br />
16. Zwemer CF, Whitesall SE, D'Alecy LG. Cardiopulmonary-cerebral resuscitation with<br />
100% oxygen exacerbates neurological dysfunction following nine minutes <strong>of</strong><br />
normothermic cardiac arrest in dogs. Resuscitation. 1994;27:159-170.<br />
17. Rubertsson S, Karlsson T, Wiklund L. Systemic oxygen uptake during experimental<br />
closed-chest cardiopulmonary resuscitation using air or pure oxygen ventilation. Acta<br />
Anaesthesiol Sc<strong>and</strong>. 1998;42:32-38.<br />
Subchapter 20.4. Perioperative Glucose Control<br />
Background<br />
Diabetes is a well-known risk factor for perioperative medical complications. Poor<br />
glucose control is an independent risk factor for surgical site infections 1-5 in a range <strong>of</strong> surgical<br />
procedures. Increased risk for infection is thought to result from a combination <strong>of</strong> clinically<br />
apparent effects <strong>of</strong> longst<strong>and</strong>ing hyperglycemia (eg, macro- <strong>and</strong> microvascular occlusive<br />
disease) <strong>and</strong> subtle immunologic defects, most notably neutrophil dysfunction. 6-12<br />
Hyperglycemia may also impair the function <strong>of</strong> complement <strong>and</strong> antibodies, reducing the<br />
opsonic potential <strong>of</strong> these factors <strong>and</strong> impairing phagocytosis, further reducing barriers to<br />
infection. 13,14 Although many <strong>of</strong> the clinically apparent manifestations <strong>of</strong> diabetes are not easily<br />
reversed in the perioperative period, there is a small literature that suggests that improving<br />
glucose control can improve immunologic function <strong>and</strong> reduce the incidence <strong>of</strong> surgical site<br />
infections (SSI). 6-8,12<br />
237
Perioperative management <strong>of</strong> glucose for diabetic patients commonly includes<br />
withholding or administering a reduced dose <strong>of</strong> the patients’ usual hypoglycemic agent(s) <strong>and</strong><br />
commencing a low-rate intravenous glucose infusion while patients are NPO prior to surgery.<br />
The infusion is continued postoperatively until the patient is able to eat <strong>and</strong> resume outpatient<br />
diabetes therapy. Often a sliding scale insulin regimen, a schedule <strong>of</strong> subcutaneous regular<br />
insulin dosage contingent on capillary blood glucose measurements, is also continued through<br />
the perioperative period. However, use <strong>of</strong> a sliding scale may result in wide variations in serum<br />
glucose, 15 opening the rationale <strong>of</strong> this method to question. 16-18<br />
Practice description<br />
Aggressive glucose control in the perioperative period can be achieved using a<br />
continuous intravenous insulin infusion (CII). Nursing staff monitor fingerstick (or arterial line<br />
drop-<strong>of</strong>-blood sample) glucose measurements <strong>and</strong> adjust the infusion rate based on a protocol<br />
intended to maintain serum glucose within a certain range. For example, the target range for the<br />
original Portl<strong>and</strong> Protocol was between 151 <strong>and</strong> 200 mg/dL. 16,19,20 In the most recent version, the<br />
range is between 125 <strong>and</strong> 175 mg/dL. 21<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Little evidence exists to describe the practice <strong>of</strong> CII in prevention <strong>of</strong> surgical site<br />
infections in broad surgical practice. The small amount <strong>of</strong> evidence available describes its use in<br />
patients undergoing cardiac surgery, primarily coronary artery bypass grafting (CABG).<br />
Diabetes is a well-described risk factor for sternal wound infections, a catastrophic complication<br />
<strong>of</strong> median sternotomy. 19,22-25 Sternal wound infections occur in 0.8% to 2% <strong>of</strong> unselected<br />
patients undergoing median sternotomy <strong>and</strong> CABG. 20,22,23 Diabetic patients, who comprise<br />
between 17 <strong>and</strong> 20% <strong>of</strong> all patients undergoing CABG, have been reported to have an incidence<br />
<strong>of</strong> sternal wound infections as high as 5.6%. 26 Such infections are associated with marked<br />
increases in morbidity <strong>and</strong> costs. Furnary et al reported that patients with sternal wound<br />
infections had an average increased length <strong>of</strong> stay <strong>of</strong> 16 days <strong>and</strong> a higher mortality rate (19%<br />
vs. 3.8% in patients without sternal wound infections). 20 (See also Subchapter 20.1).<br />
Opportunities for Impact<br />
More than 700,000 Americans underwent open-heart surgery in 1998 alone. 27 Up to 20%<br />
<strong>of</strong> these patients may be c<strong>and</strong>idates for continuous insulin infusion. Although CII is included in<br />
the recent ACC/AHA Guidelines for CABG Surgery, 28 there are no data on the extent to which<br />
the measure is currently used during cardiac or other surgical procedures.<br />
Study Designs <strong>and</strong> Outcomes<br />
We identified one prospective before-after study that compared rates <strong>of</strong> deep sternal<br />
wound infections (DSWI) in diabetic patients undergoing CABG before <strong>and</strong> after<br />
implementation <strong>of</strong> an aggressive CII protocol. 20 DSWI included infections involving the sternum<br />
or mediastinal tissues, including mediastinitis. An older study from the same authors was not<br />
reviewed as it reported findings at an earlier point in the same trial. 19 Additional studies<br />
examined the use <strong>of</strong> CII in perioperative patients but did not report Level 1 clinical outcomes<br />
relevant to patient safety (eg, mortality, wound infection) <strong>and</strong> were also not reviewed. 29<br />
238
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Furnary et al found that aggressive glucose control with CII was associated with a<br />
reduction in deep sternal wound infections. 20 The effect <strong>of</strong> the intervention remained statistically<br />
significant in a logistic regression model adjusting for multiple potential confounding variables.<br />
Furthermore, the demographic characteristics were generally biased against the CII group, which<br />
had a significantly higher percentage <strong>of</strong> patients with hypertension, renal insufficiency, <strong>and</strong><br />
obesity but fewer patients with congestive heart failure. However, the authors did not adjust for<br />
long-term markers <strong>of</strong> glucose control such as glycosylated hemoglobin, nor did they describe<br />
other changes in patient care systems that resulted from changing patients to insulin infusions.<br />
Continuous insulin infusions require closer attention by nursing staff both for monitoring <strong>of</strong><br />
infusion equipment <strong>and</strong> for frequent measurements <strong>of</strong> blood glucose. It is possible that the<br />
improved outcomes were due to closer overall attention to the patient. Although 74% <strong>of</strong> DSWI<br />
occurred after initial discharge (raising the concern that the shorter length <strong>of</strong> stay in the sliding<br />
scale insulin group may have resulted in some infection not being detected), the authors reported<br />
that they directly followed-up all diabetic patients for one year from the time <strong>of</strong> surgery. 30 The<br />
personnel, equipment, surgical techniques, <strong>and</strong> use <strong>of</strong> prophylactic antibiotics were similar<br />
throughout the study period. 31 Nonetheless, it is likely that secular trends in the care <strong>of</strong> patients<br />
undergoing cardiac surgery account for some <strong>of</strong> the impact attributed to CII.<br />
Potential for Harm<br />
Hypoglycemic episodes are the most concerning adverse event associated with intensive<br />
glucose management with intravenous insulin. These episodes result in a range <strong>of</strong> medical<br />
complications, from delirium to myocardial infarction resulting from increased sympathetic<br />
activity. Furnary noted that, using the st<strong>and</strong>ardized protocol in their study, no cases <strong>of</strong><br />
symptomatic hypoglycemia occurred in either group <strong>of</strong> patients. 30 However, CII protocols<br />
intended to maintain normoglycemia in surgical patients have been associated with high rates<br />
(40%) <strong>of</strong> postoperative hypoglycemia requiring treatment (
intervention group, the number <strong>of</strong> DSWIs prevented was 10 (95% CI: 4-21) <strong>and</strong> the average cost<br />
to prevent one DSWI was approximately $21,000 (95% CI: $10,000-$52,500). Of course, these<br />
figures do not incorporate the potential effects <strong>of</strong> the intervention on other sites <strong>of</strong> infection,<br />
mortality, adverse events, <strong>and</strong> patients’ preferences (utilities) for these possible health states.<br />
Comment<br />
An increasing body <strong>of</strong> evidence demonstrates that tight control <strong>of</strong> blood glucose<br />
improves overall outcomes <strong>of</strong> patients with diabetes. 35-37 Emerging data, coupled with an<br />
increasing appreciation <strong>of</strong> the deleterious effects <strong>of</strong> hyperglycemia on immune function, strongly<br />
support the supposition that aggressive control <strong>of</strong> perioperative glucose reduces the incidence <strong>of</strong><br />
surgical site infections. Although the practice has been implemented at a number <strong>of</strong> institutions<br />
<strong>and</strong> is also being used in diabetic patients undergoing non-cardiac surgeries, 34 studies <strong>of</strong> its<br />
effectiveness in these settings have not yet been published. Until additional evidence is available,<br />
preferably from blinded r<strong>and</strong>omized controlled trials, the intervention can be considered<br />
promising but not yet proven to be causally associated with improved outcomes.<br />
240
Table 20.4.1. Prospective, before-after study <strong>of</strong> aggressive perioperative glucose control*<br />
Study Study<br />
Population<br />
Furnary,<br />
1999 20<br />
2467 diabetic<br />
patients<br />
undergoing<br />
cardiac<br />
surgery at a<br />
community<br />
hospital<br />
Comparison Groups Results†<br />
968 patients treated with<br />
sliding scale SQ insulin<br />
(1987-91)<br />
1499 patients treated with<br />
CII to target glucose <strong>of</strong><br />
150-200 mg/dL until<br />
POD 3 (1991-97)<br />
241<br />
Deep surgical wound infections<br />
Unadjusted: 1.9% vs. 0.8% (p=0.011)<br />
Adjusted RR 0.34 (95% CI: 0.14-0.74)<br />
Mortality: 6.1% vs. 3.0% (p=0.03)<br />
Length <strong>of</strong> Stay: 10.7d vs. 8.5d (p
11. Mowat A, Baum J. Chemotaxis <strong>of</strong> polymorphonuclear leukocytes from patients with<br />
diabetes mellitus. N Engl J Med. 1971;284:621-627.<br />
12. MacRury SM, Gemmell CG, Paterson KR, MacCuish AC. Changes in phagocytic function<br />
with glycaemic control in diabetic patients. J Clin Pathol. 1989;42:1143-1147.<br />
13. Hennessey PJ, Black CT, Andrassy RJ. Nonenzymatic glycosylation <strong>of</strong> immunoglobulin G<br />
impairs complement fixation. JPEN J Parenter Enteral Nutr. 1991;15:60-64.<br />
14. Black CT, Hennessey PJ, Andrassy RJ. Short-term hyperglycemia depresses immunity<br />
through nonenzymatic glycosylation <strong>of</strong> circulating immunoglobulin. J Trauma.<br />
1990;30:830-832.<br />
15. Queale WS, Seidler AJ, Brancati FL. Glycemic control <strong>and</strong> sliding scale insulin use in<br />
medical inpatients with diabetes mellitus. Arch Intern Med. 1997;157:545-552.<br />
16. Zinman B. Insulin regimens <strong>and</strong> strategies for IDDM. Diabetes Care. 1993;16(Suppl 3):24-<br />
28.<br />
17. Kletter GG. Sliding scale fallacy. Arch Intern Med. 1998;158:1472.<br />
18. Shagan B. Does anyone here know how to make insulin work backwards? Why sliding<br />
scale insulin coverage doesn't work. Pract Diabetol. 1990;9:1-4.<br />
19. Zerr KJ, Furnary AP, Grunkemeier GL, Bookin S, Kanhere V, Starr A. Glucose control<br />
lowers the risk <strong>of</strong> wound infection in diabetics after open heart operations. Ann Thorac<br />
Surg. 1997;63:356-361.<br />
20. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous intravenous insulin infusion<br />
reduces the incidence <strong>of</strong> deep sternal wound infection in diabetic patients after cardiac<br />
surgical procedures. Ann Thorac Surg. 1999;67:352-360.<br />
21. Starr Wood Cardiac Group. Starr Wood Research - Continuous Intravenous Insulin<br />
Infusion. Available at: http://www.starrwood.com/research/insulin.html. Accessed June 11,<br />
2001.<br />
22. Borger MA, Rao V, Weisel RD, Ivanov J, Cohen G, Scully HE, et al. Deep sternal wound<br />
infection: risk factors <strong>and</strong> outcomes. Ann Thorac Surg. 1998;65:1050-1056.<br />
23. Trick WE, Scheckler WE, Tokars JI, Jones KC, Reppen ML, Smith EM, et al. Modifiable<br />
risk factors associated with deep sternal site infection after coronary artery bypass grafting.<br />
J Thorac Cardiovasc Surg. 2000;119:108-114.<br />
24. Zacharias A, Habib RH. Factors predisposing to median sternotomy complications. Deep<br />
vs superficial infection. Chest. 1996;110:1173-1178.<br />
25. Whang W, Bigger JT. Diabetes <strong>and</strong> outcomes <strong>of</strong> coronary artery bypass graft surgery in<br />
patients with severe left ventricular dysfunction: results from The CABG Patch Trial<br />
database. The CABG Patch Trial Investigators <strong>and</strong> Coordinators. J Am Coll Cardiol.<br />
2000;36:1166-1172.<br />
26. Golden SH, Peart-Vigilance C, Kao WH, Brancati FL. Perioperative glycemic control <strong>and</strong><br />
the risk <strong>of</strong> infectious complications in a cohort <strong>of</strong> adults with diabetes. Diabetes Care.<br />
1999;22:1408-1414.<br />
27. American Heart Association. 2001 Heart <strong>and</strong> Stroke Statistical Update. Available at:<br />
http://www.americanheart.org/statistics/medical.html. Accessed June 11, 2001.<br />
28. Eagle KA, Guyton RA, David<strong>of</strong>f R, Ewy GA, Fonger J, Gardner TJ, et al. ACC/AHA<br />
Guidelines for Coronary Artery Bypass Graft Surgery: A Report <strong>of</strong> the American College<br />
<strong>of</strong> Cardiology/American Heart Association Task Force on Practice Guidelines (Committee<br />
to Revise the 1991 Guidelines for Coronary Artery Bypass Graft Surgery). American<br />
College <strong>of</strong> Cardiology/American Heart Association. J Am Coll Cardiol. 1999;34:1262-<br />
1347.<br />
242
29. Lazar HL, Chipkin S, Philippides G, Bao Y, Apstein C. Glucose-insulin-potassium<br />
solutions improve outcomes in diabetics who have coronary artery operations. Ann Thorac<br />
Surg. 2000;70:145-150.<br />
30. Furnary AP. Continuous intravenous insulin infusion reduces the incidence <strong>of</strong> deep sternal<br />
wound infection in diabetic patients after cardiac surgical procedures [discussion]. Ann<br />
Thorac Surg. 1999;67:360-62.<br />
31. CTSNet, the Cardiothoracic Surgery Network. Use <strong>of</strong> historical controls may weaken<br />
conclusions. Available at: Available at: http:www.ctsnet.org/forum/78/0/2240. Accessed<br />
June 18, 2001.<br />
32. Chaney MA, Nikolov MP, Blakeman BP, Bakhos M. Attempting to maintain<br />
normoglycemia during cardiopulmonary bypass with insulin may initiate postoperative<br />
hypoglycemia. Anesth Analg. 1999;89:1091-1095.<br />
33. Star Wood Cardiac Group. The Portl<strong>and</strong> Insulin Protocol - Frequently Asked Questions<br />
Page 1. Available at: http://www.starrwood.com/research/Insulin_FAQ1.html. Accessed<br />
June 11, 2001.<br />
34. Starr Wood Cardiac Group. The Portl<strong>and</strong> Insulin Protocol - Frequently Asked Questions<br />
Page 2. Available at: http://www.starrwood.com/research/Insulin_FAQ2.html. Accessed<br />
June 11, 2001.<br />
35. The effect <strong>of</strong> intensive treatment <strong>of</strong> diabetes on the development <strong>and</strong> progression <strong>of</strong> longterm<br />
complications in insulin-dependent diabetes mellitus. The Diabetes Control <strong>and</strong><br />
Complications Trial Research Group. N Engl J Med. 1993;329:977-986.<br />
36. The effect <strong>of</strong> intensive diabetes therapy on the development <strong>and</strong> progression <strong>of</strong> neuropathy.<br />
The Diabetes Control <strong>and</strong> Complications Trial Research Group. Ann Intern Med.<br />
1995;122:561-568.<br />
37. Retinopathy <strong>and</strong> nephropathy in patients with type 1 diabetes four years after a trial <strong>of</strong><br />
intensive therapy. The Diabetes Control <strong>and</strong> Complications Trial/Epidemiology <strong>of</strong><br />
Diabetes Interventions <strong>and</strong> Complications Research Group. N Engl J Med. 2000;342:381-<br />
389.<br />
243
244
Chapter 21. Ultrasound Guidance <strong>of</strong> Central Vein Catheterization<br />
Jeffrey M. Rothschild, MD, MPH<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
The multiple indications for central venous catheters (CVCs) include parenteral nutrition,<br />
intravascular depletion, access for vasoactive medications, hemodynamic monitoring,<br />
cardiopulmonary arrest, difficult peripheral intravenous (IV) access <strong>and</strong> long-term IV access for<br />
medications, such as antibiotics. 1,2 While these catheters can be life saving, they are also<br />
associated with significant risks. 3 These risks increase in association with several characteristics,<br />
including patient anatomy (eg, morbid obesity, cachexia, or local scarring from surgery or<br />
radiation treatment), patient setting (eg, patients receiving mechanical ventilation or during<br />
emergencies such as cardiac arrest) <strong>and</strong> co-morbidities (eg, bullous emphysema or<br />
coagulopathy). 3-5<br />
CVCs are placed by clinicians whose training <strong>and</strong> experience may vary greatly. The<br />
procedure takes place in a variety <strong>of</strong> hospital settings including intensive care units, emergency<br />
departments, operating rooms, pre- <strong>and</strong> post-anesthesia care units, hemodialysis units, cardiac<br />
catheterization units <strong>and</strong> other inpatient settings. Outpatient placement <strong>of</strong> CVCs has also become<br />
commonplace, occurring in hemodialysis centers <strong>and</strong> oncology centers providing outpatient<br />
chemotherapy.<br />
Percutaneous insertions <strong>of</strong> CVCs are usually performed by “blind” techniques that rely<br />
on anatomic l<strong>and</strong>marks – ie, palpable or visible structures with known relationships to the<br />
desired vein. For example, the infraclavicular approach to the subclavian vein requires correct<br />
localization <strong>of</strong> the clavicle reference site, suprasternal notch <strong>and</strong> sternocleidomastoid-clavicular<br />
triangle l<strong>and</strong>marks, proper positioning <strong>of</strong> the patient <strong>and</strong> operator <strong>and</strong> correct venipuncture point<br />
depth, direction <strong>and</strong> insertion angle. Analogously, the various approaches to the internal jugular<br />
vein require thorough knowledge <strong>of</strong> this vein’s course in relation to the sternocleidomastoid<br />
muscle <strong>and</strong> carotid artery. 1<br />
Newer technologies, such as portable ultrasound (US) devices, provide bedside imaging<br />
<strong>of</strong> the central veins during catheter placement. The advantages associated with US guided CVC<br />
placement include detection <strong>of</strong> anatomic variations <strong>and</strong> exact vessel location, avoidance <strong>of</strong><br />
central veins with pre-existing thrombosis that may prevent successful CVC placement, <strong>and</strong><br />
guidance <strong>of</strong> both guidewire <strong>and</strong> catheter placement after initial needle insertion. This review<br />
assesses the impact <strong>of</strong> real-time ultrasound guidance on improving the safety <strong>of</strong> CVC insertions.<br />
Practice Description<br />
Real-time ultrasound guidance <strong>of</strong> CVC insertion provides the operator with visualization<br />
<strong>of</strong> the desired vein <strong>and</strong> the surrounding anatomic structures prior to <strong>and</strong> during insertion <strong>of</strong> the<br />
needle, guidewire <strong>and</strong> catheter. Previous studies <strong>of</strong> US location <strong>of</strong> vessels followed by<br />
subsequent catheter placement with l<strong>and</strong>mark techniques found no advantages over st<strong>and</strong>ard<br />
l<strong>and</strong>mark techniques. 3 Real-time US guidance, on the other h<strong>and</strong>, appears to improve the success<br />
rate <strong>and</strong> decrease the complication rate associated with CVC placement.<br />
Two types <strong>of</strong> real-time ultrasound guidance are described. The Doppler US guidance<br />
method include audio-guided Doppler, 6 fingertip pulsed Doppler 7 <strong>and</strong> probe-in-the-needle<br />
245
technology. 6,8-10 The non-Doppler US guidance methods (subsequently referred to as US<br />
guidance) includes US with needle guidance 11,12 or without needle guidance. 13-15<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The annual number <strong>of</strong> all CVCs insertions in the United States is not known, but is<br />
estimated at “several million” for subclavian-vein catheters. 3 When aggregated with the various<br />
types <strong>of</strong> catheters placed into the central venous circulation <strong>and</strong> the increasing utilization <strong>of</strong><br />
CVCs among routine surgical patients, critically ill patients (in both emergency departments <strong>and</strong><br />
intensive care units), <strong>and</strong> for the management <strong>of</strong> many patients undergoing hemodialysis or<br />
chemotherapy, the total number <strong>of</strong> CVC placements may be many times greater than estimates<br />
for subclavian CVCs alone. 3<br />
Unsuccessful insertion <strong>of</strong> CVCs may occur in up to 20% <strong>of</strong> cases. 3-5 The hazards<br />
associated with attempted CVC insertion (whether successful or not) include arterial puncture,<br />
hematoma, pneumothorax, hemothorax, chylothorax, brachial plexus injury, arrhythmias, air<br />
embolus, catheter malposition <strong>and</strong> catheter knotting. 3-5 Other complications associated with<br />
CVCs, such as infection, thrombosis, arterial-venous fistula <strong>and</strong> vascular or cardiac erosion, are<br />
not usually associated with needle insertion but occur after catheter placement. 3-5<br />
The frequency <strong>of</strong> complications associated with CVC placement is quite variable, largely<br />
due to differences among selected venous insertion sites, the degree <strong>of</strong> prior operator experience<br />
<strong>and</strong> the presence <strong>of</strong> previously described risk factors. In general, the rate <strong>of</strong> major CVC<br />
complications (eg, pneumothorax or vessel laceration requiring repair) <strong>and</strong> minor complications<br />
(eg, arterial puncture without significant hemorrhage, transient catheter malposition) is between<br />
0.5 <strong>and</strong> 10%. 3-5<br />
In addition to complications, several quality-<strong>of</strong>-care issues are associated with problems<br />
in CVC insertion. For example, a CVC insertion that requires multiple attempts may engender<br />
considerable patient anxiety <strong>and</strong> discomfort. More importantly, a prolonged insertion process<br />
may delay the infusion <strong>of</strong> life-saving fluids or medications during an emergency.<br />
Opportunities for Impact<br />
The majority <strong>of</strong> CVC insertions are placed using the l<strong>and</strong>mark method. As set forth<br />
above, the number <strong>of</strong> catheters placed annually <strong>and</strong> the proportion currently inserted without US<br />
guidance is not known. Also unknown is the proportion <strong>of</strong> catheters placed in those patients who<br />
may benefit most from the US technique—those with multiple risk factors, those in high risk<br />
settings such as the intensive care unit, or those undergoing catheter placement by inexperienced<br />
operators.<br />
There are a variety <strong>of</strong> catheters that require access to the central veins. These include<br />
single <strong>and</strong> multi-lumen CVCs, tunneled <strong>and</strong> non-tunneled catheters, <strong>and</strong> larger, more rigid<br />
catheter introducers that permit passage <strong>of</strong> thinner, more flexible devices (eg, pulmonary artery<br />
catheters, diagnostic cardiac catheters <strong>and</strong> temporary transvenous pacemakers). All centrallyplaced<br />
catheters require an initial needle insertion into the vein, followed by a guidewire to<br />
permit passage <strong>of</strong> the catheter.<br />
Study Designs<br />
One meta-analysis <strong>and</strong> 10 original studies were analyzed. Among the 10 original studies,<br />
9 were r<strong>and</strong>omized control studies <strong>and</strong> 1 study was a quasi-r<strong>and</strong>omized control trial (see Table<br />
21.1). 11 The meta-analysis 16 includes 6 references cited in this chapter. 6,8,10,12-14 Four articles<br />
cited in this chapter were not cited in the 1996 meta-analysis, including 3 that were published<br />
246
after 1996 9,15,17 <strong>and</strong> one that included a quasi-r<strong>and</strong>omized design. 17 Two studies included in the<br />
meta-analysis were not included in this chapter because they were from non-English language<br />
journals. 7,18<br />
All <strong>of</strong> the studies analyzed for this chapter were prospective, non-blinded <strong>and</strong><br />
r<strong>and</strong>omized, with the exception <strong>of</strong> a quasi-r<strong>and</strong>omized design involving alternate week<br />
allocation <strong>of</strong> patients to receive the intervention. 11 R<strong>and</strong>omization was at the patient (rather than<br />
physician) level in all studies.<br />
The 10 original studies include 5 using an ultrasound guidance technique without<br />
Doppler <strong>and</strong> 5 using the Doppler US technique. Sites <strong>of</strong> catheterization included the internal<br />
jugular (IJ) veins in 6 studies, the subclavian (SC) veins in 3 studies <strong>and</strong> the femoral veins in 1<br />
study. The study populations are diverse <strong>and</strong> include intensive care unit patients, surgical<br />
patients both preoperatively <strong>and</strong> intraoperatively, cardiac patients in both the catheterization <strong>and</strong><br />
coronary care units, emergency department patients in cardiopulmonary arrest <strong>and</strong> oncology<br />
center outpatients.<br />
Examples <strong>of</strong> studies excluded from analysis are studies lacking comparison control<br />
groups, 19-27 studies <strong>of</strong> US use but without real-time guidance, 3,28 <strong>and</strong> simulations <strong>of</strong> CVC<br />
placement rather than completed procedures. 29<br />
Study Outcomes<br />
Studies included for review reported a combination <strong>of</strong> clinical outcomes. These outcomes<br />
include Level 1 complications that resulted in increased patient morbidity (eg, pneumothorax)<br />
<strong>and</strong> Level 2 complications that represent potential adverse events (eg, unwanted arterial puncture<br />
without sequelae). Most studies also reported the number <strong>of</strong> venipuncture attempts to achieve<br />
successful CVC placement as well as the time required to complete successful CVC insertions.<br />
These are considered Level 2 outcomes because increased venipuncture attempts are associated<br />
with increased complication rates. 4<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The 1996 meta-analysis 16 estimated that real-time US guidance for CVC insertion is<br />
associated with a significant reduction in placement failures as compared with the usual<br />
l<strong>and</strong>mark techniques (relative risk 0.32, 95% CI: 0.18-0.55). In addition, this review estimated<br />
that US guidance results in decreased complications during attempted CVC placements (relative<br />
risk 0.22, 95% CI: 0.10-0.45), corresponding to a relative risk reduction <strong>of</strong> 78%. 16 The mean<br />
number <strong>of</strong> attempted venipunctures till successful CVC insertion was significantly reduced with<br />
real-time US guidance (relative risk 0.60, 95% CI: 0.45-0.79), corresponding to a relative risk<br />
reduction <strong>of</strong> 40%. 16<br />
Two <strong>of</strong> the 3 studies reviewed for this chapter <strong>and</strong> not included in the meta-analysis<br />
(because they were published subsequently) deserve mention because <strong>of</strong> contrary findings. Both<br />
1998 studies 9,17 included operators with significant prior experience placing CVCs by the usual<br />
l<strong>and</strong>mark method. The overall failure rate for both l<strong>and</strong>mark <strong>and</strong> US guidance techniques was<br />
very low in one study. 17 The other study found statistically insignificant negative results<br />
associated with US guidance, owing to a high failure rate during the initial learning period for<br />
the newly-introduced US guidance technique, <strong>and</strong> a very low (1.3%) overall complication rate<br />
for CVC placement. 9<br />
With the exception <strong>of</strong> femoral venous catheterization during cardiopulmonary<br />
resuscitation, 15 the studies reviewed for this chapter did not find reductions in insertion time<br />
when using real-time US guidance. With 2 exceptions, 910 the cited studies reached statistical<br />
247
significance for at least one <strong>of</strong> the 3 outcome measures (morbidity, potential adverse events, <strong>and</strong><br />
number <strong>of</strong> venipuncture attempts). The most favorable outcomes associated with real-time US<br />
guidance were found in studies <strong>of</strong> inexperienced operators. 6,12,13<br />
Potential for Harm<br />
The additional equipment <strong>and</strong> manipulation associated with real-time US guidance for<br />
CVC insertion may increase the rate <strong>of</strong> catheter-related infections, but published studies have not<br />
included these complications. In emergency settings, the increased length <strong>of</strong> time required to<br />
place CVCs under US guidance (usually an additional 30 seconds to several minutes) may result<br />
in unacceptable delays. Potential harmful consequences resulting from real-time US guidance for<br />
CVC placement relate to changes in training <strong>and</strong> subsequent dependence on this technology.<br />
Supporters <strong>of</strong> this technology argue that increased competence <strong>and</strong> anatomic knowledge gained<br />
with US guidance will enhance performance <strong>of</strong> customary, unaided CVC placement. It is unclear<br />
if trainees who have performed CVC placement only with US assistance will have different<br />
complication rates when placing CVCs in practice settings without US equipment. Therefore, for<br />
certification <strong>of</strong> qualification, trainees may need to demonstrate competence with both US <strong>and</strong><br />
non-US guided CVC placement.<br />
Costs <strong>and</strong> Implementation<br />
The major impediments to the widespread implementation <strong>of</strong> US guidance for CVC<br />
insertion are the purchase costs <strong>of</strong> the US machines. A typical machine costs $11,000-16,000<br />
(including the probes), with a single machine able to serve most critical care units. Depending on<br />
the layout <strong>of</strong> units placing CVCs, an average 400-bed hospital would require 1-3 machines for<br />
use outside <strong>of</strong> the operating room. (Departments <strong>of</strong> Anesthesia may require only one or two<br />
machines, as experienced anesthesiologists can continue to place most CVCs without US<br />
guidance). Hospitals in which nurses place peripherally inserted central catheter (PICC) lines<br />
using US guidance typically facilitate this function with a single machine, which could be dually<br />
used for CVCs depending on workflow <strong>and</strong> volume. In one study, the price <strong>of</strong> the Doppler-Smart<br />
needles (which are not required for non-Doppler US guidance) was $40-70 as compared with $3-<br />
5 for the st<strong>and</strong>ard needles. 9 The cost <strong>of</strong> training new operators (including those whose only prior<br />
experience is with the l<strong>and</strong>mark technique) requires further evaluation.<br />
248
Comment<br />
Real-time US guidance for CVC insertion, with or without Doppler assistance, improves<br />
catheter insertion success rates, reduces the number <strong>of</strong> venipuncture attempts prior to successful<br />
placement, <strong>and</strong> reduces the number <strong>of</strong> complications associated with catheter insertion.<br />
However, these benefits may not accrue until after the initial learning period for operators<br />
already experienced in the l<strong>and</strong>mark techniques. 9<br />
There are no studies comparing the impact <strong>of</strong> CVC insertion with US guidance on overall<br />
patient outcomes (eg, mortality, length <strong>of</strong> stay). In addition, many <strong>of</strong> the complications<br />
associated with CVC insertion are minor or easily treated. The reduction in venipuncture<br />
attempts is likely associated with reductions in the pain <strong>and</strong> discomfort associated with CVC<br />
placement, though this has not been measured.<br />
The greatest benefit <strong>of</strong> US guidance may apply to the novice or inexperienced operator<br />
<strong>and</strong> for all operators in high-risk situations. <strong>Patient</strong>s with one or more risk factors, (eg, critically<br />
ill patients on positive pressure ventilation with generalized edema <strong>and</strong> coagulopathy), may reap<br />
the greatest benefit. CVC insertion training incorporating real-time ultrasound guided techniques<br />
may provide additional valuable learning benefits for new operators. This knowledge may<br />
improve the success rate <strong>of</strong> insertion <strong>of</strong> CVCs without US guidance. Simulation training has<br />
demonstrated improved identification <strong>of</strong> the desired veins with US as compared to l<strong>and</strong>mark<br />
techniques. 29<br />
Finally, it should be noted that in addition to real-time US guidance, other approaches<br />
may reduce the risks associated with CVC insertion. PICC lines are gaining widespread<br />
acceptance <strong>and</strong> may be an acceptable substitute for CVCs for certain indications (eg, long-term<br />
IV access or parenteral nutrition). 30,31 US guidance has also been demonstrated to improve the<br />
insertion <strong>of</strong> PICCs. 32-34 Increases in the use <strong>of</strong> PICCs may help justify the purchase <strong>of</strong> ultrasound<br />
machines by individual hospitals.<br />
For patients requiring replacement <strong>of</strong> existing CVCs, guidewire exchanges <strong>of</strong>fer a way <strong>of</strong><br />
inserting new catheters without new venipuncture attempts. A systematic review has found that<br />
guidewire exchanges are associated with fewer mechanical complications than new-site<br />
replacement, but may be associated with greater risks for catheter-related infections (Chapter<br />
16). 35<br />
Alternative methods for teaching CVC insertion skills to novices (eg, first-year resident<br />
physicians <strong>and</strong> medical students) have successfully employed multidisciplinary approaches<br />
including cadaver demonstrations. 36 Future methods for teaching CVC insertion may employ<br />
computerized technologies for simulations (see also Chapter 45). Haptic, or touch-related<br />
techniques use virtual reality models to create immersive simulated environments that recreate<br />
the sensation <strong>of</strong> performing a procedure. 37,38 Through the use <strong>of</strong> this <strong>and</strong> other new technologies,<br />
novice operators may gain experience <strong>and</strong> confidence prior to clinical CVC insertion attempts<br />
<strong>and</strong> further improve patient safety.<br />
249
Table 21.1. Ultrasound <strong>and</strong> Doppler ultrasound guidance <strong>of</strong> central vein catheters*<br />
Study Setting <strong>and</strong><br />
Population<br />
Tertiary care, teaching<br />
hospital ICU 13<br />
Tertiary care, teaching<br />
hospital, CT surgical<br />
patients 14<br />
Tertiary care, teaching<br />
hospital, cardiac<br />
patients 11<br />
Urban, teaching<br />
hospital ICU 12<br />
Urban, teaching<br />
hospital ED, during<br />
CPR 15<br />
Tertiary care, teaching<br />
hospital, CT/ vascular<br />
surgery patients 8<br />
British hospital,<br />
cardiac surgery <strong>and</strong><br />
ICU patients 10<br />
Tertiary care, teaching<br />
hospital ICU <strong>and</strong> OR;<br />
high-risk patients 6<br />
French teaching<br />
hospital ICU; low-risk<br />
patients 17<br />
Tertiary care,<br />
outpatient oncology<br />
center 9<br />
Year<br />
Published<br />
Intervention Study<br />
Design,<br />
Outcomes<br />
Ultrasound<br />
1990 US guidance for IJ CVC insertion<br />
without needle guide; concurrent<br />
feedback from an US technician<br />
1991 US guidance (7.5 <strong>and</strong> 5.0 MHz<br />
transducers) for IJ CVC insertion<br />
without needle guide<br />
1993 US guidance (7.5 MHz<br />
transducer) <strong>of</strong> IJ cannulation for<br />
cardiac catheterization <strong>and</strong> CVC<br />
insertion, with needle guide<br />
1995 US guidance (7.5 MHz<br />
transducer) for SC CVC insertion<br />
with needle guide<br />
1997 US guidance (7.5 MHz<br />
transducer) for femoral CVC<br />
insertion without needle guide<br />
Doppler Ultrasound<br />
1994 Doppler US guidance <strong>of</strong> IJ CVC<br />
insertion with probe in the needle<br />
1994 Doppler US guidance <strong>of</strong> IJ CVC<br />
insertion with probe in the needle<br />
1995 Audio-guided Doppler US<br />
guidance for IJ CVC insertion<br />
with probe in the needle<br />
1998 Pulsed Doppler US guidance for<br />
SC CVC insertion without needle<br />
guide<br />
1998 Doppler US guidance for SC<br />
CVC insertion with probe in the<br />
needle<br />
250<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Level 1<br />
Level 2<br />
Relative Risk Reduction (%)†<br />
Failed<br />
Catheter<br />
Insertion<br />
100 NS<br />
Mean<br />
Insertion<br />
Attempts<br />
Required§<br />
44 NA<br />
Complications<br />
100 44 83 NS<br />
100 48 80<br />
86 48 90<br />
71 54 100<br />
0 52 0<br />
-50 NS<br />
17 NS<br />
63 18 NS<br />
-32 §<br />
-46 §<br />
0<br />
88<br />
0 67<br />
NA -53 NS<br />
* CPR indicates cardiopulmonary resuscitation; CT, cardiothoracic; ED, emergency department; ICU, intensive care<br />
unit; IJ, internaljugular vein; NA, not available; NS, not statistically significant; OR, operating room; RCT,<br />
r<strong>and</strong>omized controlled trial; SC, subclavian vein; <strong>and</strong> US, ultrasound guidance.<br />
† The percentage relative risk reduction reflects the risks associated with CVC placement <strong>and</strong> the percentage change<br />
resulting from US guidance. Negative values indicate an increase in risks associated with US guidance.<br />
§ Relative risk reduction <strong>of</strong> the insertion attempts to success reflects the relative reduction in the mean number <strong>of</strong> needle<br />
insertions attempted per patient until successful CVC placement resulting from US guidance.
References<br />
1. Civetta JM, Taylor RW, Kirby RR. <strong>Critical</strong> Care. 3rd ed. Philadelphia: Lippincott-Raven<br />
Publishers, 1997.<br />
2. Shoemaker W, Ayres S, Grenvik A, Holbrook P. Textbook <strong>of</strong> <strong>Critical</strong> Care. 4 ed.<br />
Philadelphia: WB Saunders, 2000.<br />
3. Mansfield PF, Hohn DC, Fornage BD, Gregurich MA, Ota DM. Complications <strong>and</strong> failures<br />
<strong>of</strong> subclavian-vein catheterization. N Engl J Med. 1994;331:1735-1738.<br />
4. Sznajder JI, Zveibil FR, Bitterman H, Weiner P, Bursztein S. Central vein catheterization.<br />
Failure <strong>and</strong> complication rates by three percutaneous approaches. Arch Intern Med.<br />
1986;146:259-261.<br />
5. Bernard RW, Stahl WM. Subclavian vein catheterizations: a prospective study. I. Noninfectious<br />
complications. Ann Surg. 1971;173:184-190.<br />
6. Gilbert TB, Seneff MG, Becker RB. Facilitation <strong>of</strong> internal jugular venous cannulation<br />
using an audio-guided Doppler ultrasound vascular access device: results from a<br />
prospective, dual-center, r<strong>and</strong>omized, crossover clinical study. Crit Care Med. 1995;23:60-<br />
65.<br />
7. Branger B, Zabadani B, Vecina F, Juan JM, Dauzat M. Continuous guidance for venous<br />
punctures using a new pulsed Doppler probe: efficiency, safety. Nephrologie. 1994;15:137-<br />
140.<br />
8. Gratz I, Afshar M, Kidwell P, Weiman DS, Shariff HM. Doppler-guided cannulation <strong>of</strong> the<br />
internal jugular vein: a prospective, r<strong>and</strong>omized trial. J Clin Monit. 1994;10:185-188.<br />
9. Bold RJ, Winchester DJ, Madary AR, Gregurich MA, Mansfield PF. Prospective,<br />
r<strong>and</strong>omized trial <strong>of</strong> Doppler-assisted subclavian vein catheterization. Arch Surg.<br />
1998;133:1089-1093.<br />
10. Vucevic M, Tehan B, Gamlin F, Berridge JC, Boylan M. The SMART needle. A new<br />
Doppler ultrasound-guided vascular access needle. Anaesthesia. 1994;49:889-891.<br />
11. Denys BG, Uretsky BF, Reddy PS. Ultrasound-assisted cannulation <strong>of</strong> the internal jugular<br />
vein. A prospective comparison to the external l<strong>and</strong>mark-guided technique. Circulation.<br />
1993;87:1557-1562.<br />
12. Gualtieri E, Deppe SA, Sipperly ME, Thompson DR. Subclavian venous catheterization:<br />
greater success rate for less experienced operators using ultrasound guidance. Crit Care<br />
Med. 1995;23:692-697.<br />
13. Mallory DL, McGee WT, Shawker TH, Brenner M, Bailey KR, Evans RG, et al.<br />
Ultrasound guidance improves the success rate <strong>of</strong> internal jugular vein cannulation. A<br />
prospective, r<strong>and</strong>omized trial. Chest. 1990;98:157-160.<br />
14. Troianos CA, Jobes DR, Ellison N. Ultrasound-guided cannulation <strong>of</strong> the internal jugular<br />
vein. A prospective, r<strong>and</strong>omized study. Anesth Analg. 1991;72:823-826.<br />
15. Hilty WM, Hudson PA, Levitt MA, Hall JB. Real-time ultrasound-guided femoral vein<br />
catheterization during cardiopulmonary resuscitation. Ann Emerg Med. 1997;29:331-336.<br />
16. R<strong>and</strong>olph AG, Cook DJ, Gonzales CA, Pribble CG. Ultrasound guidance for placement <strong>of</strong><br />
central venous catheters: a meta-analysis <strong>of</strong> the literature. Crit Care Med. 1996;24:2053-<br />
2058.<br />
17. Lefrant JY, Cuvillon P, Benezet JF, Dauzat M, Peray P, Saissi G, et al. Pulsed Doppler<br />
ultrasonography guidance for catheterization <strong>of</strong> the subclavian vein: a r<strong>and</strong>omized study.<br />
Anesthesiology. 1998;88:1195-1201.<br />
251
18. Scherhag A, Klein A, Jantzen J. Cannulation <strong>of</strong> the internal jugular vein using 2 ultraosnic<br />
techniques:A comparitive controlled study. Anaesthesist 1989;38:633-638.<br />
19. Fry WR, Clagett GC, O'Rourke PT. Ultrasound-guided central venous access. Arch Surg.<br />
1999;134:738-740.<br />
20. Gordon AC, Saliken JC, Johns D, Owen R, Gray RR. US-guided puncture <strong>of</strong> the internal<br />
jugular vein: complications <strong>and</strong> anatomic considerations. J Vasc Interv Radiol. 1998;9:333-<br />
338.<br />
21. Sato S, Ueno E, Toyooka H. Central venous access via the distal femoral vein using<br />
ultrasound guidance. Anesthesiology. 1998;88:838-839.<br />
22. Docktor B, So CB, Saliken JC, Gray RR. Ultrasound monitoring in cannulation <strong>of</strong> the<br />
internal jugular vein: anatomic <strong>and</strong> technical considerations. Can Assoc Radiol J.<br />
1996;47:195-201.<br />
23. Hatfield A, Bodenham A. Portable ultrasound for difficult central venous access. Br J<br />
Anaesth. 1999;82:822-826.<br />
24. Gallieni M, Cozzolino M. Uncomplicated central vein catheterization <strong>of</strong> high risk patients<br />
with real time ultrasound guidance. Int J Artif Organs. 1995;18:117-121.<br />
25. Geddes CC, Walbaum D, Fox JG, Mactier RA. Insertion <strong>of</strong> internal jugular temporary<br />
hemodialysis cannulae by direct ultrasound guidance—a prospective comparison <strong>of</strong><br />
experienced <strong>and</strong> inexperienced operators. Clin Nephrol. 1998;50:320-325.<br />
26. Cavatorta F, Zollo A, Campisi S, Trabassi E, De Lucia E, Galli S, et al. Internal jugular<br />
vein catheterization under echographic guidance. Int J Artif Organs. 1993;16:820-822.<br />
27. Hrics P, Wilber S, Bl<strong>and</strong>a MP, Gallo U. Ultrasound-assisted internal jugular vein<br />
catheterization in the ED. Am J Emerg Med. 1998;16:401-403.<br />
28. Hayashi H, Tsuzuku M, Amano M. Simplified echo-guided internal jugular vein puncture:<br />
a comparison to the l<strong>and</strong>mark-guided technique. Anesth.Analg. 1998;86:SCA89<br />
29. Hirvela E, Parsa C, Aalmi O, Kelly E, Goldstein L. Skills <strong>and</strong> risk assessment <strong>of</strong> central<br />
line placement using bedside simulation with 3-dimensional ultrasound guidance system.<br />
Crit Care Med. 2000;28:A78-A78<br />
30. Lam S, Scannell R, Roessler D, Smith MA. Peripherally inserted central catheters in an<br />
acute-care hospital. Arch Intern Med. 1994;154:1833-1837.<br />
31. Smith JR, MD, Friedell ML, Cheatham ML, Martin SP, Cohen MJ, et al. Peripherally<br />
Inserted Central Catheters Revisited. Am J Surg. 1998;176:208-211.<br />
32. LaRue GD,. Efficacy <strong>of</strong> Ultrasonography in Peripheral Venous Cannulation. J Intraven<br />
Nurs. 2000; 23:29-34.<br />
33. S<strong>of</strong>ocleous CT, Schur I, Cooper SG, Quintas JC, Brody L, Shelin R. Sonographically<br />
guided placement <strong>of</strong> peripherally inserted central venous catheters: review <strong>of</strong> 355<br />
procedures. AJR. 1998;170:1613-1616.<br />
34. Chrisman HBM, Omary RAM, Nemcek AA, Jr, MD, Vogelzang RLM. Peripherally<br />
Inserted Central Catheters: Guidance With Ultrasound Versus Venography in 2650<br />
<strong>Patient</strong>s. J Vasc Interv Radiol. 1998;9:173-174.<br />
35. Cook D, R<strong>and</strong>olph A, Kernerman P, Cupido C, King D, Soukup C, et al. Central venous<br />
catheter replacement strategies: a systematic review <strong>of</strong> the literature. Crit Care Med.<br />
1997;25:1417-1424.<br />
36. Nip IL, Haruno MM. A systematic approach to teaching insertion <strong>of</strong> a central venous line.<br />
Acad Med. 2000;75:552.<br />
37. Hiemenz L, Stredney D, Schmalbrock P. Development <strong>of</strong> the force-feedback model for an<br />
epidural needle insertion simulator. Stud Health Technol Inform. 1998;50:272-277.<br />
252
38. Kaufmann C, Rhee P, Burris D. Telepresence surgery system enhances medical student<br />
surgery training. Stud Health Technol Inform. 1999;62:174-178.<br />
253
254
Chapter 22. The Retained Surgical Sponge<br />
Verna C. Gibbs, MD<br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Although less likely to garner public notoriety, errors relating to the failure to remove<br />
surgical instruments at the end <strong>of</strong> a procedure, (ie, needles, knife blades, electrosurgical adapters<br />
<strong>and</strong> safety pins) or sponges (known as gossypiboma; gossypium: Latin, cotton; boma: Swahili,<br />
place <strong>of</strong> concealment) are no less egregious than the better known mishaps such as “wrong-site<br />
surgery” (see Subchapter 43.2).<br />
Retained materials may cause an acute foreign body reaction with local or systemic signs<br />
that prompt investigation <strong>and</strong> reoperation. Alternatively, a fibrinous response may be elicited,<br />
<strong>and</strong> the retained instrument or sponge may become apparent some time after the original surgical<br />
procedure either serendipitously, or via fistulization into local structures. 1 The medical literature<br />
is scattered with reports <strong>of</strong> presentations <strong>of</strong> retained sponges found days, months, or even years<br />
after the original surgery. 2-5 While many cases <strong>of</strong> retained foreign body do not cause harm, some<br />
clearly do. Nevertheless, the Joint Commission on Accreditation for Healthcare Organization’s<br />
(JCAHO) sentinel event policy specifically mentions that “unintentionally retained foreign body<br />
without major permanent loss <strong>of</strong> function” do not require reporting. 6 Although JCAHO’s<br />
decision suggests that it considers these events less egregious than reportable sentinel events (eg,<br />
wrong patient surgery), retained foreign body events are far more common. This chapter reviews<br />
the problem <strong>and</strong> the scanty literature regarding safety practices to reduce the incidence <strong>of</strong><br />
retained sponges <strong>and</strong> instruments.<br />
Practice Description<br />
Surgeons <strong>and</strong> operating room teams rely upon the practice <strong>of</strong> sponge, sharp <strong>and</strong><br />
instrument counts as a means to eliminate retained surgical instruments. Counts are also a<br />
method <strong>of</strong> infection control <strong>and</strong> inventory control, <strong>and</strong> a means to prevent injury from<br />
contaminated sharps <strong>and</strong> instruments. Four separate counts have been recommended 7 : the first<br />
when the instruments are set up or the sponges unpackaged, a second before the surgical<br />
procedure begins, a third as closure begins, <strong>and</strong> the final count performed during subcuticular or<br />
skin closure.<br />
Use <strong>of</strong> this simple preventative measure is not universal. In fact, the process by which<br />
counts are performed is not st<strong>and</strong>ardized <strong>and</strong> is <strong>of</strong>ten modified according to individual hospital<br />
policy. Even when present, counts are frequently omitted or abbreviated in emergency or<br />
transvaginal surgeries, or for vaginal deliveries. 8 An adjunctive procedure to the count, used<br />
when the count could delay care <strong>and</strong> jeopardize patients’ lives or when an incorrect count is<br />
established, is an x-ray examination to detect radiopaque objects. 1,7 Since this practice is not<br />
routinely used it will not be discussed here.<br />
255
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
A literature search revealed few data to describe population or even hospital-level<br />
information regarding the prevalence <strong>of</strong> retained surgical materials. One study from a medical<br />
malpractice insurance company reported 40 cases in a 7-year period, 9 or about 1% <strong>of</strong> all claims.<br />
Because this estimate is based on malpractice insurance claims, it is sure to be a gross<br />
underestimate <strong>of</strong> the actual incidence. A recent unstructured review cited “a prevalence ranging<br />
from 1/100 to 1/5000,” <strong>and</strong> an associated mortality ranging from 11 to 35%, citing non-English<br />
language medical references. 1 Other reports are based on case series or descriptions <strong>of</strong> unusual<br />
presentations, as described above. Surgeons may not report these events for a variety <strong>of</strong> reasons,<br />
not the least <strong>of</strong> which is fear <strong>of</strong> litigation<br />
Opportunities for Impact<br />
Without accurate prevalence information, the true magnitude <strong>of</strong> the opportunity for<br />
impact is unclear.<br />
Study Designs <strong>and</strong> Outcomes<br />
Only one study provided even indirect evidence <strong>of</strong> the effectiveness <strong>of</strong> sponge <strong>and</strong><br />
instrument counts. Kaiser et al, using a retrospective review <strong>of</strong> medical malpractice claims data<br />
from a statewide insurer in Massachusetts, reviewed 67 cases where retained sponges or surgical<br />
materials were the primary reason for the claim. 9 This study is a case series without any controls<br />
(Level 4 design, Level 2 outcomes) which reported only the outcome <strong>of</strong> retained sponges, rather<br />
than the clinical consequences <strong>of</strong> these errors.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The Kaiser et al study reported that 55% <strong>of</strong> retained sponges were found after abdominal<br />
surgery <strong>and</strong> 16% after vaginal delivery. In cases with retained sponges, sponge counts had been<br />
falsely correct in 76% <strong>of</strong> non-vaginal surgeries; in 10% no sponge count had been performed at<br />
all. Falsely correct sponge counts were attributed to team fatigue, difficult operations, sponges<br />
“sticking together,” or a poor counting system. Incorrect sponge counts that were accepted prior<br />
to closure resulted from either surgeons’ dismissing the incorrect count without re-exploring the<br />
wound, or nursing staff allowing an incorrect count to be accepted. Interestingly, in 3 <strong>of</strong> 29 cases<br />
in which intraoperative x-rays were used to detect radiopaque sponges, the radiograph was<br />
falsely negative. 9<br />
Comment<br />
Although literature describing the incidence <strong>of</strong> iatrogenic foreign bodies is highly limited<br />
in quantity <strong>and</strong> quality, it is unlikely that these events are as rare as other iatrogenic<br />
complications that have drawn considerable national attention. The existing system <strong>of</strong> sponge<br />
<strong>and</strong> instrument counts probably works well, but we have no evidence to describe its actual<br />
failure rate. The little existing evidence suggests that it fails due to human-related factors (ie, the<br />
count is not performed, or is ignored, <strong>and</strong> that ancillary methods such as x-rays are also fallible.<br />
Although some have advocated CT or ultrasonography as additional methods to reduce rates <strong>of</strong><br />
these adverse events, it is possible that other technologies (eg, inventory control devices used in<br />
retail stores <strong>and</strong> libraries, possibly including bar coding (Subchapter 45.1)) may prove to be<br />
useful adjuncts. However, there are obvious logistical challenges that make such technologies<br />
too impractical at the present time. For now we are left with a paucity <strong>of</strong> data regarding the<br />
256
prevalence <strong>of</strong> this error <strong>and</strong> the effectiveness <strong>of</strong> preventative measures. Use <strong>of</strong> anonymous<br />
reporting systems may reduce the fear <strong>of</strong> litigation associated with iatrogenic foreign bodies, <strong>and</strong><br />
may allow for more accurate assessment <strong>of</strong> the incidence <strong>and</strong> causes <strong>of</strong> these events.<br />
References<br />
1. Lauwers PR, Van Hee RH. Intraperitoneal gossypibomas: the need to count sponges.<br />
World J Surg. 2000;24:521-527.<br />
2. Wig JD, Goenka MK, Suri S, Sudhakar PJ, Vaiphei K. Retained surgical sponge: an<br />
unusual cause <strong>of</strong> intestinal obstruction. J Clin Gastroenterol. 1997;24:57-58.<br />
3. Scott WW, Beall DP, Wheeler PS. The retained intrapericardial sponge: value <strong>of</strong> the lateral<br />
chest radiograph. AJR Am J Roentgenol. 1998;171:595-597.<br />
4. Sinha SK, Gupta S, Behra A, et al. Retained surgical sponge: an unusual cause <strong>of</strong><br />
malabsorption. Trop Gastroenterol. 1999;20:42-44.<br />
5. Burrel M, Capurro S, Arguis P, Vilana R. Sonographic appearance <strong>of</strong> a retained surgical<br />
sponge in the neck. J Clin Ultrasound. 2000;28:311-313.<br />
6. The Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Alert.<br />
Accreditation Committee Approves Examples Of Voluntarily Reportable Sentinel Events.<br />
May 11, 1998. Available at: http://www.jcaho.org/edu_pub/sealert/sea4.html. Accessed<br />
June 23, 2001.<br />
7. Recommended practices for sponge, sharp, <strong>and</strong> instrument counts. AORN Recommended<br />
<strong>Practices</strong> Committee. Association <strong>of</strong> periOperative Registered Nurses. AORN J.<br />
1999;70:1083-9.<br />
8. Horty J. Nurses, surgeon not negligent in retained instrument suit. Kissinger v Turner. OR<br />
Manager. 1995;11:26, 30.<br />
9. Kaiser CW, Friedman S, Spurling KP, Slowick T, Kaiser HA. The retained surgical<br />
sponge. Ann Surg. 1996;224:79-84.<br />
257
258
Chapter 23. Pre-Anesthesia Checklists To Improve <strong>Patient</strong> <strong>Safety</strong><br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Harvey J. Murff, MD<br />
Harvard <strong>Medical</strong> School<br />
Salim D. Islam, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
No matter how rote the task or how vigilant the anesthesiologist, “slips” <strong>and</strong> other errors<br />
represent expected aspects <strong>of</strong> human performance. 1 Evaluation <strong>and</strong> subsequent improvement <strong>of</strong><br />
st<strong>and</strong>ard checkout procedures promises to increase patient safety in the perioperative period by<br />
removing more <strong>of</strong> the “human factors” so <strong>of</strong>ten implicated in anesthesia adverse events. 2 The use<br />
<strong>of</strong> pre-flight checklists has been considered a key method in improving airline safety, largely due<br />
to the regular systematizing <strong>of</strong> complex procedures, <strong>and</strong> improvement <strong>of</strong> team dynamics through<br />
authority-neutral tasks. A checklist system has been proposed as part <strong>of</strong> routine pre-anesthesia<br />
care, with the American Society <strong>of</strong> Anesthesiologists <strong>and</strong> the US Food <strong>and</strong> Drug Administration<br />
(FDA) issuing general guidelines supporting checklists in 1986. 3 Subsequently, anesthesia<br />
pr<strong>of</strong>essional societies in Great Britain <strong>and</strong> Europe adopted similar st<strong>and</strong>ards. 4<br />
Practice Description<br />
In 1987, the FDA published the “Anesthesia Apparatus Checkout Recommendations” in<br />
the Federal Register (February, 1987). This “checkout list” provides practitioners with a<br />
st<strong>and</strong>ardized approach to checking anesthetic equipment prior to its use in order to ensure that<br />
the delivery system is correctly connected, adjusted, <strong>and</strong> functional. The original checklist<br />
included 24 specific processes to be performed as an initial checkout at the beginning <strong>of</strong> each<br />
day; 11 are performed between cases <strong>and</strong> after initial equipment evaluation. 5 Many clinicians<br />
regarded these protocols as too long <strong>and</strong> complex for routine use. Parties involved in revising the<br />
recommendations agreed that the average clinician should be able to check an anesthesia<br />
machine in 5 minutes or less (not possible with the 1987 recommendations). The revised<br />
checklist included only 14 major processes, 9 <strong>of</strong> which can be omitted or substantially<br />
abbreviated when the anesthesia provider uses the same equipment in successive cases. 6 The<br />
revised recommendations are available online. 7<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The earliest observational studies <strong>of</strong> mishaps within anesthesia found that equipment<br />
failure was the cause <strong>of</strong> 14% <strong>of</strong> anesthesia critical incidents. 8 Subsequent studies reduced this<br />
estimate considerably. Although equipment failure is now implicated in only 4% <strong>of</strong> anesthesia<br />
adverse events, 22% are related to failing to check equipment adequately. 2 Equipment failures<br />
can result in delivery <strong>of</strong> hypoxic gas mixtures or excessive doses <strong>of</strong> inhalational agent, or<br />
hypoventilation due to ventilator failure. These situations can be catastrophic if unrecognized,<br />
<strong>and</strong> even if recognized may result in significant morbidity (eg, delayed extubation, stroke, or<br />
myocardial infarction). 2,9<br />
259
Opportunities for Impact<br />
The FDA checkout list is considered a template which local users are “encouraged to<br />
modify to accommodate differences in equipment <strong>and</strong> variations in local clinical practice.”<br />
Estimating the frequency with which it or other checklists are used (in modified or unmodified<br />
forms) in anesthesia practice would be speculative, 5 but a survey <strong>of</strong> 4 states suggests that usage<br />
is minimal. 3 There are 41.5 million inpatient procedures each year requiring anesthesia in the<br />
US. Probably about half <strong>of</strong> these involve general anesthesia. Therefore, although the frequency<br />
<strong>of</strong> equipment failure is low, any improvement in safety from anesthesia preoperative checklists<br />
could have substantial impact.<br />
Study Designs<br />
Using a structured MEDLINE search, we identified 15 articles that discussed anesthesia<br />
checklists. Of these, only 2 studies came close to meeting our inclusion criteria (Chapter 3). 3,10<br />
Because no other studies could be found, we abstracted these studies <strong>and</strong> review them here<br />
(Table 23.1).<br />
The first investigation evaluated the ability <strong>of</strong> participating providers to detect<br />
st<strong>and</strong>ardized equipment faults using their own checking methods compared with the FDA<br />
checkout list. 3 Although the study involved a prospective design, it does not fit neatly into our<br />
classification system because the participants (anesthesiology residents <strong>and</strong> practitioners) served<br />
as their own controls. The second study 10 involved the revised FDA checkout list, <strong>and</strong> its design<br />
was modeled on the previous study.<br />
Study Outcomes<br />
The first study’s outcome <strong>of</strong> interest was the detection <strong>of</strong> 4 st<strong>and</strong>ardized equipment faults<br />
created by the investigators (Level 2). 3 The second study used similarly designed outcomes<br />
(simulated equipment faults), but defined detection <strong>of</strong> at least 50% <strong>of</strong> these faults as the primary<br />
outcome (Level 2). 10 Neither study directly connected the use <strong>of</strong> pre-anesthesia checklists to<br />
patient outcomes, although inadequate preanesthesia equipment checks have been implicated in<br />
adverse events related to equipment failures, as discussed above. 2<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
March et al 3 found that the FDA checklist was more likely to detect faults with the<br />
nitrous oxide system (65% vs. 43%, p
to anesthesia providers’ own methods, the observed utility <strong>of</strong> this practice still is likely to be<br />
quite low.<br />
Potential for Harm<br />
We were unable to find literature that implicated checklists as causes <strong>of</strong> adverse events.<br />
There exists a theoretical possibility <strong>of</strong> delays due to complex checklists, but these have not been<br />
borne out in any published studies. There has also been concern raised about compulsory<br />
checklist procedures as an unnecessary duplication <strong>of</strong> work already performed by operating<br />
room technical personnel.<br />
Costs <strong>and</strong> Implementation<br />
The FDA checklist, or local variants there<strong>of</strong>, is widely implemented <strong>and</strong> inexpensive.<br />
Although several authors have mentioned that checkout processes are not used in 100% <strong>of</strong> cases,<br />
this does not seem to reflect problems with checklists themselves. 4,12,13 The work <strong>of</strong> March <strong>and</strong><br />
Crowley 3 suggests that checklists training may be critically important. However, the most<br />
important barrier to implementation is the heterogeneity <strong>of</strong> anesthesia delivery devices<br />
themselves, which makes creation <strong>of</strong> a single, effective, broadly generalizable checklist difficult<br />
if not impossible.<br />
Comment<br />
Given its face validity <strong>and</strong> the theoretical connection between anesthesia checklists <strong>and</strong><br />
those used so effectively in aviation, preanesthesia checklists represent plausible safety tools.<br />
However, the reliability <strong>of</strong> modern anesthesia machines has reduced the frequency <strong>of</strong> mechanical<br />
failure to such a degree that adverse outcomes due to anesthesia machine failure are exceedingly<br />
rare. The checklists examined in the cited studies only examine the integrity <strong>of</strong> the anesthesia<br />
machine <strong>and</strong> ventilatory monitors, not cardiovascular monitors, airway equipment, intravenous<br />
apparatus, infusion pumps, or medications. St<strong>and</strong>ardized checklists for these other critical<br />
components <strong>of</strong> anesthesia care do not exist in the literature. This may explain the paucity <strong>of</strong><br />
literature exploring the use <strong>of</strong> checklists <strong>and</strong> their real effects on patient outcomes. Furthermore,<br />
the inability <strong>of</strong> anesthesiologists to detect preset faults in the cited studies may simply reflect the<br />
infrequency with which such faults are encountered in modern anesthesia practice.<br />
The face validity <strong>of</strong> checklists <strong>and</strong> the difficulty <strong>of</strong> “probing” their value in clinical<br />
studies make additional “pro<strong>of</strong> <strong>of</strong> concept” studies unlikely. Although future investigations could<br />
not ethically study the anesthesia machine checklist per se, they could seek to determine more<br />
effective methods for its implementation, or could develop additional checklists with a broader<br />
scope. The little evidence we have been able to uncover suggests that, like the use <strong>of</strong> voluntary<br />
guidelines elsewhere in medicine (Chapter 51), checklists are not used uniformly. This may<br />
result from a sense that these checks are low-yield, redundant, onerous, or all <strong>of</strong> the above.<br />
The need for effective checkout procedures is likely to grow as the complexity <strong>of</strong><br />
anesthesia equipment increases. This will increase the need to make checklists more sensitive<br />
<strong>and</strong> specific in detecting faults, while improving usability. These worthy goals may then serve as<br />
templates for other technologically dependent medical specialties, such as cardiac<br />
electrophysiology, <strong>and</strong> interventional radiology.<br />
261
Table 23.1. Evaluations <strong>of</strong> the FDA “checkout list” for preoperative anesthesia equipment<br />
assessment<br />
Study Study Design,<br />
Outcomes<br />
March, 1991 3 : Exposure <strong>of</strong> a total <strong>of</strong><br />
188 anesthesiology residents <strong>and</strong><br />
practitioners from multiple sites to a<br />
“mobile anesthesia study center” with 2<br />
anesthesia machines pre-set to one <strong>of</strong><br />
two different “fault sets.” Each <strong>of</strong> these<br />
sets included 4 independent machine<br />
faults. Practitioners were instructed to<br />
use their own checkout methods to<br />
assess a machine with one <strong>of</strong> the “fault<br />
sets.” After completion <strong>of</strong> this<br />
assessment, the machine was adjusted to<br />
display the other fault set, <strong>and</strong> the<br />
participant invited to use the FDA<br />
checkout list in the second assessment.<br />
Manley, 1996 9 : Similar study to above,<br />
but involving only 22 participants (a<br />
mixture <strong>of</strong> anesthesiologist, nurse<br />
anesthetists, <strong>and</strong> senior nurse anesthetist<br />
students) from only one site.<br />
Mixture <strong>of</strong><br />
Levels 2 & 3,<br />
Level 2<br />
262<br />
Results<br />
Participants detected an average <strong>of</strong><br />
only 1 in 4 machine faults. A<br />
statistically significant<br />
improvement in the fault detection<br />
rate with use <strong>of</strong> the FDA checkout<br />
compared with the participants’<br />
individual methods was observed<br />
for only 1 <strong>of</strong> the 4 fault types<br />
(involving the oxygen/nitrous<br />
oxide ratio).<br />
As above For both <strong>of</strong> the fault sets,<br />
approximately half <strong>of</strong> the<br />
participants detected fewer than<br />
50% <strong>of</strong> the 4 target faults using<br />
their own individual methods. The<br />
detection rates using the revised<br />
FDA checkout list did not differ<br />
significantly (p=0.48) from these<br />
results.
References<br />
1. Reason JT. Human Error. New York: Cambridge Univ Press; 1990.<br />
2. Cooper JB, Newbower RS, Kitz RJ. An analysis <strong>of</strong> major errors <strong>and</strong> equipment failures in<br />
anesthesia management: considerations for prevention <strong>and</strong> detection. Anesthesiology.<br />
1984;60:34-42.<br />
3. March MG, Crowley JJ. An evaluation <strong>of</strong> anesthesiologists' present checkout methods <strong>and</strong><br />
the validity <strong>of</strong> the FDA checklist. Anesthesiology. 1991;75:724-729.<br />
4. Charlton JE. Checklists <strong>and</strong> patient safety. Anaesthesia. 1990;45:425-426.<br />
5. Walker JS, Wilson M. Clinical risk management in anaesthesia. Qual Health Care.<br />
1995;4:115-121.<br />
6. Anesthesia Apparatus Checkout Recommendations, 1993. Fed Regist. 1994;59:35373-74.<br />
7. U.S. Food <strong>and</strong> Drug Adminstration. Anesthesia apparatus checkout recommendations.<br />
Available at: http://www.fda.gov/cdrh/humfac/anesckot.pdf. Accessed June 10, 2001.<br />
8. Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study<br />
<strong>of</strong> human factors. Anesthesiology. 1978;49:399-406.<br />
9. Spooner RB, Kirby RR. Equipment-related anesthetic incidents. Int Anesthesiol Clin.<br />
1984;22:133-147.<br />
10. Manley R, Cuddeford JD. An assessment <strong>of</strong> the effectiveness <strong>of</strong> the revised FDA checklist.<br />
AANA J. 1996;64:277-282.<br />
11. Degani A, Wiener EL. Cockpit checklists - concepts, design, <strong>and</strong> use. Human Factors.<br />
1993;35:345-359.<br />
12. Cooper JB, Cullen DJ, Eichhorn JH, Philip JH, Holzman RS. Administrative guidelines for<br />
response to an adverse anesthesia event. The Risk Management Committee <strong>of</strong> the Harvard<br />
<strong>Medical</strong> School's Department <strong>of</strong> Anaesthesia. J Clin Anesth. 1993;5:79-84.<br />
13. Currie M. A prospective survey <strong>of</strong> anaesthetic critical events in a teaching hospital.<br />
Anaesth Intensive Care. 1989;17:403-411.<br />
263
264
Chapter 24. The Impact Of Intraoperative Monitoring On <strong>Patient</strong> <strong>Safety</strong><br />
Salim D. Islam, MD<br />
Andrew D. Auerbach, MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Until the 1960s, intraoperative monitoring consisted <strong>of</strong> a blood pressure cuff,<br />
electrocardiogram (ECG), stethoscope, <strong>and</strong> the vigilance <strong>of</strong> the anesthesiologist. Over the next 2<br />
decades, the array <strong>of</strong> available monitors burgeoned, <strong>and</strong> clinical practice varied widely. In 1986,<br />
in an effort to improve patient safety, st<strong>and</strong>ards for intraoperative monitoring were developed<br />
<strong>and</strong> implemented by the American Society <strong>of</strong> Anesthesiologists (ASA). 1 They have been almost<br />
universally adopted by anesthesia providers in the United States <strong>and</strong> now form the st<strong>and</strong>ard <strong>of</strong><br />
care in this country. The ASA st<strong>and</strong>ards are summarized in Figure 24.1.<br />
Concurrently with the implementation <strong>of</strong> better monitoring, anesthesia-related mortality<br />
has fallen sharply. Proponents <strong>of</strong> monitoring claim that better monitoring is the reason for<br />
improvement in patient safety. 2-4 Others have claimed that advances in knowledge <strong>and</strong> training<br />
combined with the development <strong>of</strong> safer medications have had as much impact on patient safety<br />
as the adoption <strong>of</strong> monitoring st<strong>and</strong>ards. 5,6 In this chapter, we evaluate the evidence linking<br />
intraoperative monitoring to patient safety.<br />
Practice Description<br />
Intraoperative monitoring involves the use <strong>of</strong> mechanical devices to record <strong>and</strong> display<br />
physiologic parameters such as heart rate, blood pressure, oxygen saturation, <strong>and</strong> temperature.<br />
St<strong>and</strong>ard routine monitoring is noninvasive, employing blood pressure cuff, ECG, <strong>and</strong> pulse<br />
oximetry.<br />
Invasive monitors such as arterial <strong>and</strong> central venous catheters <strong>and</strong> transesophageal<br />
echocardiography may provide more detailed <strong>and</strong> timely physiologic information, but also pose<br />
an increased risk for iatrogenic complications. In practice these monitors are used selectively,<br />
<strong>and</strong> are not reviewed here.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Death due to anesthesia has become rare. In one large Canadian study involving 27,184<br />
inpatients who underwent anesthesia, physician review <strong>of</strong> 115 r<strong>and</strong>omly selected “major events”<br />
classified less than 20% as having any anesthetic involvement, with no deaths even partially<br />
attributed to anesthesia. 7 In the United States, the mortality due to general anesthesia has been<br />
estimated at approximately 5000 deaths per year (in the 1970s), 8 with approximately half that<br />
number estimated in the 1980s. 9 Thus, general anesthesia represents the one aspect <strong>of</strong> health care<br />
where the risk <strong>of</strong> death is low enough to rival the safety record achieved in other high-risk<br />
industries such as aviation. 10<br />
By contrast, morbid events (complications) related to anesthetic care are likely more<br />
prevalent <strong>and</strong> difficult to classify as preventable or unavoidable. Because certain aspects <strong>of</strong><br />
monitoring may reduce the incidence <strong>of</strong> morbid events unrelated to anesthesia, assessing the<br />
impact <strong>of</strong> monitoring practices solely on anesthetic outcomes may be inappropriate. For<br />
example, detection <strong>of</strong> a consistent decrease in intraoperative blood pressure may signal<br />
unrecognized bleeding, allowing the anesthesiologist to alert the surgeon to this possibility <strong>and</strong><br />
prompting appropriate management. While intraoperative hemorrhage does not represent an<br />
265
anesthetic complication, intraoperative blood pressure monitoring can clearly contribute to the<br />
overall safety <strong>of</strong> the surgical patient. Thus, the scope <strong>of</strong> intraoperative morbidity targeted by<br />
anesthetic monitoring practices is much broader than the set <strong>of</strong> possible complications<br />
attributable solely to the administration <strong>of</strong> anesthesia. 7-9<br />
Opportunities for Impact<br />
In the United States, there are no m<strong>and</strong>atory regulations for monitoring practices.<br />
However, virtually all anesthesiologists abide by the monitoring st<strong>and</strong>ards set forth by the 1986<br />
ASA st<strong>and</strong>ards, last modified in 1998 (Figure 24.1). Although these st<strong>and</strong>ards were implemented<br />
with only speculative evidence <strong>of</strong> their benefit, 4 few clinicians doubt their merit.<br />
Study Designs<br />
Using a structured MEDLINE search, we identified articles presenting data related to the<br />
impact <strong>of</strong> perioperative monitoring. Many <strong>of</strong> these studies 11-17 involved Level 4 designs (eg,<br />
observational studies without a control group). For instance, several <strong>of</strong> the articles 11-13,15 reported<br />
data from the Australian Incident Monitoring Study <strong>and</strong> involved analysis <strong>of</strong> a case series <strong>of</strong><br />
2000 incident reports without accompanying controls. Other studies only indirectly pertained to<br />
intraoperative monitoring. One study surveyed anesthesiologists regarding their views on the<br />
appropriate alarm settings for intraoperative blood pressure monitoring. 18 Another focused on the<br />
personnel performing intraoperative monitoring—physician anesthesiologists versus certified<br />
nurse anesthetists. 19 We chose not to purse this contentious <strong>and</strong> intensely political comparison,<br />
as few studies have compared the outcomes achieved by these two groups. Moreover, our<br />
reviewer team did not include a nurse anesthetist, making any conclusions drawn more<br />
susceptible to bias. Of the 3 remaining studies, one involved a non-r<strong>and</strong>omized clinical trial<br />
(Level 2), but a Level 3 outcome. 20<br />
The remaining 2 studies met our inclusion criteria (Chapter 3). One was a retrospective<br />
analysis <strong>of</strong> anesthesia accidents before <strong>and</strong> after the implementation <strong>of</strong> monitoring st<strong>and</strong>ards<br />
(Level 3), 2 <strong>and</strong> the other used a prospective, r<strong>and</strong>omized, controlled trial design (Level 1) to<br />
assess the impact <strong>of</strong> pulse oximetry on postoperative complications (Level 1 outcome). 21<br />
Study Outcomes<br />
The 2 studies 2,21 meeting the methodologic inclusion criterion reported morbidity <strong>and</strong><br />
mortality (Level 1) attributable to anesthesia, ie, a major complication or death occurring in the<br />
immediate postoperative period not obviously explained by the patient’s underlying condition or<br />
the operation itself.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Through a review <strong>of</strong> cases reported to a liability insurer, Eichhorn identified 11 major<br />
intraoperative accidents solely attributable to anesthesia among over 1,000,000 cases performed<br />
at the nine Harvard hospitals from 1976-1988. 2 Eight <strong>of</strong> these accidents were judged to be<br />
preventable as they were caused by failure to ventilate or to deliver adequate oxygen to the<br />
patient. Only one <strong>of</strong> these accidents occurred after the adoption <strong>of</strong> monitoring st<strong>and</strong>ards in mid-<br />
1985, supporting the safety benefit <strong>of</strong> intraoperative monitoring st<strong>and</strong>ards, although the<br />
difference in accident frequency was not statistically significant.<br />
In a multicenter, r<strong>and</strong>omized, controlled trial <strong>of</strong> 20,802 surgical patients, Moller et al 21<br />
studied the impact <strong>of</strong> perioperative pulse oximetry on patient outcome. Despite the large sample,<br />
the authors were unable to show a difference in in-hospital mortality or postoperative<br />
266
complications. During anesthesia <strong>and</strong> in the post-anesthesia care unit (PACU), more episodes <strong>of</strong><br />
hypoxemia <strong>and</strong> myocardial ischemia were detected in patients monitored with pulse oximetry. 21<br />
Potential for Harm<br />
Routine noninvasive monitoring carries minimal (although not zero) additional risk for<br />
iatrogenic complications from the devices themselves. Current st<strong>and</strong>ard <strong>of</strong> practice requires that<br />
they be used in all cases <strong>of</strong> general or regional anesthesia. However the number <strong>of</strong> monitors <strong>and</strong><br />
their concomitant alarms raises the possibility <strong>of</strong> additional harm. A study <strong>of</strong> monitor alarms in<br />
the intensive care unit (ICU) suggested that monitor alarms might actually reduce quality <strong>of</strong> care<br />
because <strong>of</strong> their high frequency <strong>and</strong> low specificity. In this study, an alarm occurred every 37<br />
minutes, <strong>and</strong> in the majority <strong>of</strong> cases (72%) no change in management was indicated as a<br />
result. 22<br />
Costs <strong>and</strong> Implementation<br />
The costs <strong>of</strong> intraoperative monitors are largely fixed in the acquisition cost <strong>of</strong> the<br />
monitoring device. Incremental patient costs for disposables are minimal.<br />
Comment<br />
The inability <strong>of</strong> a very large multicenter study 21 to detect a benefit in morbidity <strong>and</strong><br />
mortality from pulse oximetry—by all accounts the most useful monitor—suggests that the<br />
magnitude <strong>of</strong> benefit may be so small that an adequate study to detect this difference may not be<br />
feasible. Along with capnography (carbon dioxide monitoring), pulse oximetry is <strong>of</strong>ten cited as<br />
the monitoring method most able to detect potential critical incidents early enough to prevent<br />
adverse outcomes. 2,6 This conjecture is supported by the ASA Closed Claims Study. Analyzing<br />
1175 claims, the study concluded that the combination <strong>of</strong> pulse oximetry <strong>and</strong> capnography<br />
“could be expected” to help prevent anesthetic-related morbidity <strong>and</strong> mortality. 23<br />
Despite a lack <strong>of</strong> r<strong>and</strong>omized trial data, the practice <strong>of</strong> noninvasive intraoperative<br />
monitoring has become st<strong>and</strong>ard <strong>of</strong> care. This has resulted from the ASA Monitoring St<strong>and</strong>ards<br />
<strong>and</strong> physicians' faith in the practice based on its face value, along with some confirmatory<br />
evidence drawn from incident reporting systems. 11,16,17 As such, it seems likely that future<br />
research into intraoperative monitoring will be unable to study approaches that do not include<br />
st<strong>and</strong>ard, noninvasive monitoring. Future investigation might seek to determine which<br />
monitoring methods detect “near misses” more effectively.<br />
Moving beyond non-invasive techniques, there is a great need to identify which<br />
specialized monitors provide a safety benefit in selected patient populations. The use <strong>of</strong><br />
pulmonary artery catheters for monitoring critically ill patients represents a well-known example<br />
<strong>of</strong> a practice with substantial face validity but unclear impact on patient outcomes. 24,25 In<br />
addition, new, noninvasive alternatives to invasive monitors (eg, esophageal or spirometry-based<br />
cardiac output monitors) may ultimately allow us to obtain the same information at less risk to<br />
the patient.<br />
267
Figure 24.1. ASA st<strong>and</strong>ards for basic anesthetic monitoring*<br />
St<strong>and</strong>ard 1: Qualified anesthesia personnel shall be present in the room<br />
throughout the conduct <strong>of</strong> all general anesthetics, regional anesthetics <strong>and</strong><br />
monitored anesthesia care<br />
St<strong>and</strong>ard 2: During all anesthetics, the patient’s oxygenation, ventilation,<br />
circulation, <strong>and</strong> temperature shall be continually* evaluated<br />
Oxygenation:<br />
Oxygen analyzer for inspired gases<br />
Observation <strong>of</strong> the patient<br />
Pulse oximetry<br />
Ventilation:<br />
Auscultation <strong>of</strong> breath sounds<br />
Observation <strong>of</strong> the patient<br />
Observation <strong>of</strong> the reservoir bag<br />
Capnography (Carbon dioxide monitoring)<br />
Circulation:<br />
Continuous* ECG display<br />
Heart rate <strong>and</strong> BP recorded every 5 minutes<br />
Evaluation <strong>of</strong> circulation<br />
Auscultation <strong>of</strong> heart sounds<br />
Palpation <strong>of</strong> pulse<br />
Pulse plethysmography<br />
Pulse oximetry<br />
Intraarterial pressure tracing<br />
Temperature:<br />
Monitor temperature when changes are intended, anticipated, or suspected<br />
*The term “continuous” means prolonged without interruption; "continually"<br />
means repeated regularly <strong>and</strong> frequently. ECG indicates electrocardiography;<br />
BP, blood pressure.<br />
268
References<br />
1. American Society <strong>of</strong> Anesthesiologists. St<strong>and</strong>ards <strong>of</strong> the American Society <strong>of</strong><br />
Anesthesiologists: St<strong>and</strong>ards for Basic Anesthetic Monitoring. Available at:<br />
http://www.asahq.org/St<strong>and</strong>ards/02.html. Accessed May 23, 2001.<br />
2. Eichhorn JH. Prevention <strong>of</strong> intraoperative anesthesia accidents <strong>and</strong> related severe injury<br />
through safety monitoring. Anesthesiology. 1989;70:572-577.<br />
3. Eichhorn JH. Effect <strong>of</strong> monitoring st<strong>and</strong>ards on anesthesia outcome. Int Anesthesiol Clin.<br />
1993;31:181-196.<br />
4. Keenan RL, Boyan CP. Cardiac arrest due to anesthesia. A study <strong>of</strong> incidence <strong>and</strong> causes.<br />
JAMA. 1985;253:2373-2377.<br />
5. Orkin FK. Practice st<strong>and</strong>ards: the Midas touch or the emperor's new clothes?<br />
Anesthesiology. 1989;70:567-571.<br />
6. Brodsky JB. What intraoperative monitoring makes sense? Chest. 1999;115:101S-105S.<br />
7. Cohen MM, Duncan PG, Pope WD, Biehl D, Tweed WA, MacWilliam L, et al. The<br />
Canadian four-centre study <strong>of</strong> anaesthetic outcomes: II. Can outcomes be used to assess the<br />
quality <strong>of</strong> anaesthesia care? Can J Anaesth. 1992;39:430-439.<br />
8. Phillips OC, Capizzi LS. Anesthesia mortality. Clin Anesth. 1974;10:220-244.<br />
9. Deaths during general anesthesia: technology-related, due to human error, or unavoidable?<br />
An ECRI technology assessment. J Health Care Technol. 1985;1:155-175.<br />
10. National Transportation <strong>Safety</strong> Board. Aviation Accident Statistics. Available at:<br />
http://www.ntsb.gov/aviation/Stats.htm. Accessed May 28, 2001.<br />
11. Webb RK, van der Walt JH, Runciman WB, Williamson JA, Cockings J, Russell WJ, et al.<br />
The Australian Incident Monitoring Study. Which monitor? An analysis <strong>of</strong> 2000 incident<br />
reports. Anaesth Intensive Care. 1993;21:529-542.<br />
12. Klepper ID, Webb RK, Van der Walt JH, Ludbrook GL, Cockings J. The Australian<br />
Incident Monitoring Study. The stethoscope: applications <strong>and</strong> limitations–an analysis <strong>of</strong><br />
2000 incident reports. Anaesth Intensive Care. 1993;21:575-578.<br />
13. Cockings JG, Webb RK, Klepper ID, Currie M, Morgan C. The Australian Incident<br />
Monitoring Study. Blood pressure monitoring–applications <strong>and</strong> limitations: an analysis <strong>of</strong><br />
2000 incident reports. Anaesth Intensive Care. 1993;21:565-569.<br />
14. Hewer I, Drew B, Karp K, Stotts N. The utilization <strong>of</strong> automated ST segment analysis in<br />
the determination <strong>of</strong> myocardial ischemia. AANA J. 1997;65:351-356.<br />
15. Williamson JA, Webb RK, Cockings J, Morgan C. The Australian Incident Monitoring<br />
Study. The capnograph: applications <strong>and</strong> limitations–an analysis <strong>of</strong> 2000 incident reports.<br />
Anaesth Intensive Care. 1993;21:551-557.<br />
16. Spittal MJ, Findlay GP, Spencer I. A prospective analysis <strong>of</strong> critical incidents attributable<br />
to anaesthesia. Int J Qual Health Care. 1995;7:363-371.<br />
17. Findlay GP, Spittal MJ, Radcliffe JJ. The recognition <strong>of</strong> critical incidents: quantification <strong>of</strong><br />
monitor effectiveness. Anaesthesia. 1998;53:595-598.<br />
18. Asbury AJ, Rolly G. Theatre monitor alarm settings: a pilot survey in Scotl<strong>and</strong> <strong>and</strong><br />
Belgium. Anaesthesia. 1999;54:176-180.<br />
19. Silber JH, Kennedy SK, Even-Shoshan O, Chen W, Koziol LF, Showan AM, et al.<br />
Anesthesiologist direction <strong>and</strong> patient outcomes. Anesthesiology. 2000;93:152-163.<br />
20. Kay J, Neal M. Effect <strong>of</strong> automatic blood pressure devices on vigilance <strong>of</strong> anesthesia<br />
residents. J Clin Monit. 1986;2:148-150.<br />
269
21. Moller JT, Johannessen NW, Espersen K, Ravlo O, Pedersen BD, Jensen PF, et al.<br />
R<strong>and</strong>omized evaluation <strong>of</strong> pulse oximetry in 20,802 patients: II. Perioperative events <strong>and</strong><br />
postoperative complications. Anesthesiology. 1993;78:445-453.<br />
22. Chambrin MC, Ravaux P, Calvelo-Aros D, Jaborska A, Chopin C, Boniface B.<br />
Multicentric study <strong>of</strong> monitoring alarms in the adult intensive care unit (ICU): a descriptive<br />
analysis. Intensive Care Med. 1999;25:1360-1366.<br />
23. Tinker JH, Dull DL, Caplan RA, Ward RJ, Cheney FW. Role <strong>of</strong> monitoring devices in<br />
prevention <strong>of</strong> anesthetic mishaps: a closed claims analysis. Anesthesiology. 1989;71:541-<br />
546.<br />
24. Connors AF, Sper<strong>of</strong>f T, Dawson NV, Thomas C, Harrell FE, Wagner D, et al. The<br />
effectiveness <strong>of</strong> right heart catheterization in the initial care <strong>of</strong> critically ill patients.<br />
SUPPORT Investigators. JAMA. 1996;276:889-897.<br />
25. Ivanov R, Allen J, Calvin JE. The incidence <strong>of</strong> major morbidity in critically ill patients<br />
managed with pulmonary artery catheters: a meta-analysis. Crit Care Med. 2000;28:615-<br />
619.<br />
270
Chapter 25. Beta-blockers <strong>and</strong> Reduction <strong>of</strong> Perioperative Cardiac Events<br />
Andrew D. Auerbach MD, MPH<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
As the most common complications <strong>of</strong> major noncardiac surgery, myocardial infarction<br />
<strong>and</strong> cardiovascular death have long been a focus <strong>of</strong> preoperative evaluations 1-4 <strong>and</strong> a target <strong>of</strong><br />
perioperative management strategies. Until recently, methods to reduce the incidence <strong>of</strong> these<br />
complications depended upon preoperative assessments <strong>of</strong> risk combining clinical evaluation<br />
with clinical prediction rules, followed by additional tests or revascularization procedures, as<br />
appropriate. 1 The benefit <strong>of</strong> preoperative revascularization remains unclear, as no r<strong>and</strong>omized<br />
prospective trial has demonstrated its benefit. 5 Indeed, concern exists that preoperative<br />
intervention might prove detrimental, as the net benefit in terms <strong>of</strong> reduced perioperative cardiac<br />
events may be <strong>of</strong>fset by the risks <strong>of</strong> the revascularization strategy itself. Newer strategies,<br />
including the use <strong>of</strong> percutaneous transluminal angioplasty as the revascularization modality,<br />
may have promise. 6 Large prospective trials examining these approaches are underway. 5<br />
Strong evidence links myocardial ischemia with postoperative myocardial events. 7,8 One<br />
study found postoperative ischemia increased the odds <strong>of</strong> postoperative myocardial events 21fold.<br />
9 Based on findings from observational studies that beta-blockade blunts<br />
electrocardiographic signs <strong>of</strong> ischemia, 10-12 recent trials have examined the effects <strong>of</strong><br />
perioperative beta-blocker administration on patient outcomes. Results <strong>of</strong> these investigations<br />
are extremely promising, <strong>and</strong> beta-blockade may represent an important new method <strong>of</strong> reducing<br />
perioperative cardiac risk. This chapter reviews the evidence from r<strong>and</strong>omized controlled trials<br />
examining the effect <strong>of</strong> perioperative beta-blockade on cardiac events (ie, myocardial ischemia,<br />
angina, myocardial infarction, pulmonary edema, <strong>and</strong> cardiac death).<br />
Practice Description<br />
Although published studies have employed different agents, doses, <strong>and</strong> dosing schedules,<br />
the general approach in each study has been similar: administration <strong>of</strong> a therapeutic dose <strong>of</strong> betablocker<br />
prior to induction <strong>of</strong> anesthesia, followed by beta-blockade through the operation <strong>and</strong> in<br />
the postoperative period. In all regimens, the dose is titrated to a target heart rate, generally 70<br />
beats per minute or lower.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Myocardial cardiac events are the most common medical complication <strong>of</strong> surgery,<br />
occurring in 2-5% <strong>of</strong> patients undergoing non-cardiac surgery 13 <strong>and</strong> as many as 30% <strong>of</strong> patients<br />
undergoing vascular surgery. 14,15 Perioperative cardiac events are associated with a mortality rate<br />
<strong>of</strong> nearly 60% per event, 14,16 prolonged hospitalization, <strong>and</strong> higher costs. 16,17 The prevalence <strong>of</strong><br />
these events <strong>and</strong> their high mortality have made the prevention <strong>of</strong> perioperative cardiac ischemia<br />
the subject <strong>of</strong> practice guidelines 1,17 <strong>and</strong> numerous prediction rules 13,16,18 to detect patients at<br />
high risk for these complications.<br />
271
Opportunities for Impact<br />
As a relatively new therapy, few data describe the use <strong>of</strong> perioperative beta-blockade in<br />
clinical practice. However, evidence suggests it is utilized infrequently. A recent observational<br />
study in the Netherl<strong>and</strong>s <strong>of</strong> patients undergoing vascular surgery showed that only 27% <strong>of</strong> these<br />
high-risk patients received beta-blockers perioperatively. 19<br />
Study Designs<br />
Using a structured MEDLINE search, we identified 4 relevant r<strong>and</strong>omized controlled<br />
trials <strong>of</strong> the effectiveness <strong>of</strong> perioperative beta-blockade in reducing perioperative cardiac<br />
events, including myocardial ischemia <strong>and</strong> cardiac or all-cause mortality (Table 25.1). A<br />
r<strong>and</strong>omized trial by Harwood et al was excluded because both groups received beta-blockers (ie,<br />
there was no control group). 20 Although data from a study by Wallace et al 21 were derived from<br />
one <strong>of</strong> the r<strong>and</strong>omized trials included in this review, 22 it reported effects <strong>of</strong> beta-blockade upon<br />
different outcomes (ie, myocardial ischemia) <strong>and</strong> was included in our review. There was<br />
sufficient evidence available to limit the review to studies <strong>of</strong> Level 1 design. Observational<br />
studies, such as those by Pasternack et al <strong>and</strong> Boersma et al, 11,19 are not included.<br />
Study Outcomes<br />
The studies identified included a range <strong>of</strong> clinical outcomes: 2 included assessment <strong>of</strong><br />
myocardial ischemia 12,23 <strong>and</strong> 3 reported myocardial infarction, pulmonary edema, cardiac death,<br />
or all-cause mortality (Level 1 outcomes). 15,22,23<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Of studies reporting the effect <strong>of</strong> beta-blockers on perioperative ischemia (Level 2<br />
outcome), all but one found a statistically significant reduction in ischemia among treated<br />
patients. Wallace et al, 21 in a subset analysis <strong>of</strong> data from Mangano et al, 24 reported less frequent<br />
perioperative myocardial ischemia in atenolol-treated patients. Stone et al 25 suggested a similar<br />
effect <strong>of</strong> beta-blockade on Holter-monitor documented myocardial ischemia. However, the<br />
authors did not report the types <strong>of</strong> procedures included in their sample, nor did they statistically<br />
compare baseline patient characteristics, leaving their conclusions open to debate. Raby et al 12<br />
also found a significant beneficial effect <strong>of</strong> beta-blockade using a continuous infusion <strong>of</strong> esmolol<br />
in high-risk patients undergoing vascular surgery. Although Urban et al also found a reduction in<br />
perioperative ischemia, this difference failed to reach statistical significance. 23 These findings<br />
may be explained in part by a relatively low cardiac risk in Urban’s cohort, who were<br />
undergoing elective total knee replacement. The patients in many <strong>of</strong> the other studies were at<br />
higher risk <strong>of</strong> cardiac events, as demonstrated by rates <strong>of</strong> ischemia in the control groups. In<br />
studies finding a statistical difference, rates <strong>of</strong> ischemia were between 28% <strong>and</strong> 73% in controls,<br />
as compared with the 15% rate <strong>of</strong> ischemia observed in Urban's control group.<br />
Of studies reporting cardiac events <strong>and</strong> cardiac mortality, 2 reported significant<br />
improvement in patient outcomes due to beta-blockade. In a study <strong>of</strong> male veterans undergoing<br />
major noncardiac surgery, Mangano et al 22 reported a relative reduction in all-cause mortality <strong>of</strong><br />
nearly 55% at 2 years. This difference, which appeared within the first 8 months <strong>of</strong> follow-up,<br />
was ascribed to a marked reduction in cardiac events in the first year <strong>of</strong> therapy (67% reduction<br />
at year 1, 48% at year 2). However, patients in the beta-blocker group had less coronary disease<br />
at study entry, were on ACE-inhibitors more frequently, <strong>and</strong> were less likely to have betablockers<br />
discontinued perioperatively, perhaps biasing results in favor <strong>of</strong> the treatment group. 26,<br />
272
27<br />
Accounting for these differences in multivariate models <strong>of</strong> varying stringency did not<br />
invalidate their findings. 24 Although questions remain about the generalizability <strong>of</strong> results to<br />
other patient populations, the authors favored broader use <strong>of</strong> beta-blockade in the setting <strong>of</strong><br />
clinical trials.<br />
Poldermans et al 15 suggested an even greater benefit <strong>of</strong> beta-blockade among high-risk<br />
patients. These investigators enrolled patients undergoing vascular surgery who had myocardial<br />
ischemia documented by dobutamine echocardiography, with an estimated rate <strong>of</strong> perioperative<br />
cardiac event <strong>of</strong> 28%. The entire patient cohort experienced a 90% reduction in cardiac death or<br />
non-fatal myocardial infarction at 30 days. Follow-up care did not include additional therapy (ie,<br />
cardiac catheterization, revascularization), raising concerns that the research algorithm did not<br />
reflect optimal clinical practice. 28, 29 However, if the true rate <strong>of</strong> events in treated patients is low<br />
(the point estimate from this small study was 3.4%), the risks associated with revascularization 30<br />
may outweigh any benefit.<br />
In contrast to the previous 2 studies, Urban et al 23 found no statistically significant<br />
difference in rates <strong>of</strong> in-hospital myocardial infarction. It is likely that these investigators' ability<br />
to detect a difference was limited in part by the relatively small sample size <strong>and</strong> shorter length <strong>of</strong><br />
follow-up. Other studies <strong>of</strong> perioperative beta-blockade employed longer periods <strong>of</strong> follow-up to<br />
detect events up to 2 years following surgery.<br />
Differences in absolute magnitude <strong>of</strong> benefit can be ascribed in part to the cardiac risks<br />
<strong>of</strong> the patients enrolled (again reflected in event rates in the control groups in each study), with<br />
the most powerful benefits seen in patients at highest risk. The greater benefit seen in<br />
Poldermans et al’s study 15 may also be due to the fact that the study did not enroll patients who<br />
were receiving beta-blockers. <strong>Patient</strong>s who are beta-blocker naïve may have a different response<br />
to perioperative use <strong>of</strong> bisoprolol, or the preexisting use <strong>of</strong> these agents may represent a<br />
confounding factor not completely accounted for in other studies <strong>of</strong> perioperative beta-blockade.<br />
Beta-blockade may have additional beneficial effects for elderly patients. <strong>Patient</strong>s who<br />
received beta-blockers were extubated more quickly, required less medication for pain, <strong>and</strong> were<br />
alert sooner after surgery. 31 Although the unblinded nature <strong>of</strong> this study leaves its findings open<br />
to debate, the possibility <strong>of</strong> additional benefits is tantalizing <strong>and</strong> worthy <strong>of</strong> further investigation.<br />
Potential for Harm<br />
Stone et al reported high rates <strong>of</strong> bradycardia (21/89 patients) in beta-blocker treated<br />
patients, “half” <strong>of</strong> whom required atropine therapy. However, the vague descriptions <strong>and</strong> more<br />
general problems with the study's design make it difficult to interpret the significance <strong>of</strong> these<br />
events in clinical practice. Adverse events related to the use <strong>of</strong> beta-blockers in other reviewed<br />
studies were infrequent (10% or less in Mangano et al 22 ) <strong>and</strong> did not require therapy or result in<br />
withdrawal <strong>of</strong> the medication. Similar rates <strong>of</strong> side effects have been noted in studies examining<br />
beta-blockade in patients undergoing cardiac surgery. 20,32,33 One study examining use <strong>of</strong><br />
propranolol in patients undergoing thoracotomy for pneumonectomy suggested that patients<br />
receiving beta-blockers had twice the rate <strong>of</strong> postoperative congestive heart failure (4/50 vs.<br />
8/50, p
Costs <strong>and</strong> Implementation<br />
The costs <strong>of</strong> beta-blockers are generally low, <strong>and</strong> the systems required to use them<br />
according to the protocols used in these studies are already in place. In addition, there is the<br />
potential for significant cost-savings if routine use <strong>of</strong> beta-blockers allows a safe reduction in the<br />
use <strong>of</strong> extensive preoperative cardiovascular testing.<br />
Comment<br />
Results from several well-designed clinical trials suggest that use <strong>of</strong> beta-blockers in the<br />
perioperative period is associated with significant reductions in patient cardiac morbidity <strong>and</strong><br />
mortality. In the future such therapy may reduce the need for additional tests <strong>and</strong><br />
revascularization procedures, 14 further reducing costs <strong>of</strong> care. However, several questions<br />
regarding its use remain, <strong>and</strong> should be topics <strong>of</strong> future research.<br />
First, no clear data suggest an advantage <strong>of</strong> one particular beta-blocking agent over<br />
another. Studies to date have employed several different beta-blockers, suggesting that the<br />
efficacy <strong>of</strong> beta-blockade is class dependent if titrated to physiologically active dosages. Other<br />
(alpha-1 selective) sympatholytics also improve patient outcomes, 36 raising the possibility that<br />
combined alpha-beta antagonists (ie, labetolol) may have benefit. Second, results from<br />
Shammash et al document the hazards <strong>of</strong> discontinuation <strong>of</strong> beta-blockers immediately<br />
postoperatively, 35 <strong>and</strong> most protocols employed treatment regimens that extended longer - even<br />
up to one month following surgery. The current studies suggest beta-blockade should be<br />
continued for at least one week postoperatively. Third, effectiveness <strong>of</strong> beta-blockade in patients<br />
at high risk due to aortic stenosis or unstable or severe cardiovascular symptoms (New York<br />
Heart Association Class III-IV) is unknown, as these patients were not included in the reviewed<br />
studies. Similarly, its utility - both in terms <strong>of</strong> cardiac events <strong>and</strong> cost - in patients with very low<br />
risk <strong>of</strong> perioperative cardiac events (ie, those undergoing same-day or outpatient surgery,<br />
ophthalmic surgery, or those who have minimal cardiac risk) is unclear. Beta-blockade has not<br />
been studied in patients undergoing regional anesthesia or conscious sedation. In addition, no<br />
study to date has examined the use <strong>of</strong> beta-blockade in patients who have poor functional status<br />
<strong>and</strong> might otherwise be referred for additional non-invasive testing. 1,5,14,17<br />
Finally, the increasing popularity <strong>of</strong> perioperative beta-blockade, particularly catalyzed<br />
by the results <strong>of</strong> the study by Poldermans et al, 15 calls into question whether risk stratification<br />
using published guidelines or risk indices is still necessary. 28 Although beta-blockade is likely to<br />
be effective in many patients, the identification <strong>of</strong> patients at highest risk is still important, as<br />
these patients may require additional testing <strong>and</strong> therapy. A recent study <strong>of</strong> beta-blockade noted<br />
improved outcomes across a spectrum <strong>of</strong> predicted cardiac risk, but noted that cardiac events<br />
could be further reduced in high-risk patients through use <strong>of</strong> additional non-invasive testing <strong>and</strong><br />
subsequent “usual care.” 19 Thus, although beta-blockade may increase the threshold at which<br />
clinicians refer patients for additional testing, the era <strong>of</strong> risk stratification is not over.<br />
274
The use <strong>of</strong> beta-blockers to reduce perioperative cardiac events <strong>and</strong> mortality represents a<br />
major advance in perioperative medicine for some patients at intermediate <strong>and</strong> high risk for<br />
cardiac events during noncardiac surgery. Wider use <strong>of</strong> this therapy should be promoted <strong>and</strong><br />
studied, with future research focused on fine-tuning dosages <strong>and</strong> schedules <strong>and</strong> identifying<br />
populations <strong>of</strong> patients in which its use is cost-effective.<br />
275
Table 25.1. R<strong>and</strong>omized controlled trials <strong>of</strong> the effectiveness <strong>of</strong> perioperative beta-blockade*<br />
Study Participants Regimen Results† Side Effects Comments<br />
Mangano,<br />
1996 22<br />
Wallace,<br />
1998 21<br />
Polderman<br />
s, 1999 15<br />
Raby,<br />
1999 12<br />
Stone,<br />
1988 37<br />
200 patients<br />
undergoing<br />
elective<br />
noncardiac<br />
surgery<br />
112 patients with<br />
positive results on<br />
dobutamine<br />
echocardiography<br />
undergoing<br />
elective<br />
abdominal aortic<br />
or infrainguinal<br />
arterial<br />
reconstruction<br />
26 patients with<br />
preoperative<br />
ischemia by<br />
Holter monitor<br />
undergoing aortic<br />
aneurysm repair,<br />
infrainguinal<br />
arterial bypass, or<br />
carotid<br />
endarterectomy<br />
128 untreated<br />
hypertensive<br />
patients<br />
undergoing<br />
elective surgery.<br />
Hypertension<br />
Atenolol 5-10g IV<br />
30 min before entry<br />
into OR, after<br />
surgery, <strong>and</strong> 50-<br />
100g qd through<br />
hospital stay (up to<br />
7 days); Target HR<br />
55-65 bpm; doses<br />
held if HR
Urban,<br />
2000 23<br />
defined as systolic<br />
blood pressure<br />
160-200 mmHg,<br />
diastolic 90-100<br />
mmHg<br />
120 patients<br />
undergoing<br />
elective total knee<br />
arthroplasty<br />
mg po given before<br />
induction <strong>of</strong><br />
anesthesia<br />
Esmolol IV within<br />
1 hr after surgery,<br />
titrate to HR
10. Smulyan H, Weinberg SE, Howanitz PJ. Continuous propranolol infusion following<br />
abdominal surgery. JAMA. 1982;247:2539-2542.<br />
11. Pasternack PF, Grossi EA, Baumann FG, et al. Beta blockade to decrease silent myocardial<br />
ischemia during peripheral vascular surgery. Am J Surg. 1989;158:113-116.<br />
12. Raby KE, Brull SJ, Timimi F, et al. The effect <strong>of</strong> heart rate control on myocardial ischemia<br />
among high-risk patients after vascular surgery. Anesth Analg. 1999;88:477-482.<br />
13. Lee TH, Marcantonio ER, Mangione CM, et al. Derivation <strong>and</strong> prospective validation <strong>of</strong> a<br />
simple index for prediction <strong>of</strong> cardiac risk <strong>of</strong> major noncardiac surgery. Circulation.<br />
1999;100:1043-1049.<br />
14. Lee TH. Reducing cardiac risk in noncardiac surgery. N Engl J Med. 1999;341:1838-1840.<br />
15. Poldermans D, Boersma E, Bax JJ, et al. The effect <strong>of</strong> bisoprolol on perioperative mortality<br />
<strong>and</strong> myocardial infarction in high-risk patients undergoing vascular surgery. Dutch<br />
Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography Study<br />
Group. N Engl J Med. 1999;341:1789-1794.<br />
16. Goldman L. Multifactorial index <strong>of</strong> cardiac risk in noncardiac surgery: ten-year status<br />
report. J Cardiothorac Anesth. 1987;1:237-244.<br />
17. Eagle KA, Brundage BH, Chaitman BR, et al. Guidelines for perioperative cardiovascular<br />
evaluation for noncardiac surgery: an abridged version <strong>of</strong> the report <strong>of</strong> the American<br />
College <strong>of</strong> Cardiology/American Heart Association Task Force on Practice Guidelines.<br />
Mayo Clin Proc.1997;72:524-531.<br />
18. Detsky AS, Abrams HB, Forbath N, Scott JG, Hilliard JR. Cardiac assessment for patients<br />
undergoing noncardiac surgery. A multifactorial clinical risk index. Arch Intern Med.<br />
1986;146:2131-2134.<br />
19. Boersma E, Poldermans D, Bax JJ, et al. Predictors <strong>of</strong> cardiac events after major vascular<br />
surgery: Role <strong>of</strong> clinical characteristics, dobutamine echocardiography, <strong>and</strong> beta-blocker<br />
therapy. JAMA. 2001;285:1865-1873.<br />
20. Harwood TN, Butterworth J, Prielipp RC, et al. The safety <strong>and</strong> effectiveness <strong>of</strong> esmolol in<br />
the perioperative period in patients undergoing abdominal aortic surgery. J Cardiothorac<br />
Vasc Anesth. 1999;13:555-561.<br />
21. Wallace A, Layug B, Tateo I, et al. Prophylactic atenolol reduces postoperative myocardial<br />
ischemia. McSPI Research Group. Anesthesiology. 1998;88:7-17.<br />
22. Mangano DT, Layug EL, Wallace A, Tateo I. Effect <strong>of</strong> atenolol on mortality <strong>and</strong><br />
cardiovascular morbidity after noncardiac surgery. Multicenter Study <strong>of</strong> Perioperative<br />
Ischemia Research Group. N Engl J Med. 1996;335:1713-1720.<br />
23. Urban MK, Markowitz SM, Gordon MA, Urquhart BL, Kligfield P. Postoperative<br />
prophylactic administration <strong>of</strong> beta-adrenergic blockers in patients at risk for myocardial<br />
ischemia. Anesth Analg. 2000;90:1257-1261.<br />
24. Mangano DT. Effect <strong>of</strong> atenolol on mortality <strong>and</strong> cardiovascular morbidity after noncardiac<br />
surgery. N Engl J Med. 1997;336:1452.<br />
25. Stone JG, Foex P, Sear JW, Johnson LL, Khambatta HJ, Triner L. Myocardial ischemia in<br />
untreated hypertensive patients: effect <strong>of</strong> a single small oral dose <strong>of</strong> a beta-adrenergic<br />
blocking agent. Anesthesiology. 1988;68:495-500.<br />
26. Reis SE, Feldman AH. Effect <strong>of</strong> atenolol on mortality <strong>and</strong> cardiovascular morbidity after<br />
noncardiac surgery. N Engl J Med. 1997;336:1453.<br />
27. Petros JA. Effect <strong>of</strong> atenolol on mortality <strong>and</strong> cardiovascular morbidity after noncardiac<br />
surgery. N Engl J Med. 1997;336:1452.<br />
278
28. Litwack R, Gilligan D, DeGruttola V. Beta-Blockade for patients undergoing vascular<br />
surgery. N Engl J Med. 2000;342:1051-1053.<br />
29. Feldman T, Fusman B, McKinsey JF. Beta-Blockade for patients undergoing vascular<br />
surgery. N Engl J Med. 2000;342:1051-1052.<br />
30. Poldermans D, Boersma E. Beta-Blockade for patients undergoing vascular surgery. N<br />
Engl J Med. 2000;342:1052-1053.<br />
31. Zaugg M, Tagliente T, Lucchinetti E, et al. Beneficial effects from beta-adrenergic<br />
blockade in elderly patients undergoing noncardiac surgery. Anesthesiology. 1999;91:1674-<br />
1686.<br />
32. Hammon JW, Wood AJ, Prager RL, Wood M, Muirhead J, Bender HW. Perioperative beta<br />
blockade with propranolol: reduction in myocardial oxygen dem<strong>and</strong>s <strong>and</strong> incidence <strong>of</strong><br />
atrial <strong>and</strong> ventricular arrhythmias. Ann Thorac Surg. 1984;38:363-367.<br />
33. Lamb RK, Prabhakar G, Thorpe JA, Smith S, Norton R, Dyde JA. The use <strong>of</strong> atenolol in<br />
the prevention <strong>of</strong> supraventricular arrhythmias following coronary artery surgery. Eur<br />
Heart J. 1988;9:32-36.<br />
34. Bayliff CD, Massel DR, Inculet RI, et al. Propranolol for the prevention <strong>of</strong> postoperative<br />
arrhythmias in general thoracic surgery. Ann Thorac Surg. 1999;67:182-186.<br />
35. Shammash JB, Trost JC, Gold JM, Berlin JA, Golden MA, Kimmel SE. Perioperative betablocker<br />
withdrawal <strong>and</strong> mortality in vascular surgical patients. Am Heart J. 2001;141:148-<br />
153.<br />
36. Oliver MF, Goldman L, Julian DG, Holme I. Effect <strong>of</strong> mivazerol on perioperative cardiac<br />
complications during non-cardiac surgery in patients with coronary heart disease: the<br />
European Mivazerol Trial (EMIT). Anesthesiology. 1999;91:951-961.<br />
37. Stone JG, Foex P, Sear JW, Johnson LL, Khambatta HJ, Triner L. Risk <strong>of</strong> myocardial<br />
ischaemia during anaesthesia in treated <strong>and</strong> untreated hypertensive patients. Br J Anaesth.<br />
1988;61:675-679.<br />
279
Section D. <strong>Safety</strong> <strong>Practices</strong> for Hospitalized or Institutionalized Elders<br />
Chapter 26. Prevention <strong>of</strong> Falls in Hospitalized <strong>and</strong> Institutionalized Older People<br />
Subchapter 26.1. Identification Bracelets for High-Risk <strong>Patient</strong>s<br />
Subchapter 26.2. Interventions that Decrease the Use <strong>of</strong> Physical Restraints<br />
Subchapter 26.3. Bed Alarms<br />
Subchapter 26.4. Special Hospital Flooring Materials to Reduce Injuries from<br />
<strong>Patient</strong> Falls<br />
Subchapter 26.5. Hip Protectors to Prevent Hip Fracture<br />
Chapter 27. Prevention <strong>of</strong> Pressure Ulcers in Older <strong>Patient</strong>s<br />
Chapter 28. Prevention <strong>of</strong> Delirium in Older Hospitalized <strong>Patient</strong>s<br />
Chapter 29. Multidisciplinary Geriatric Consultation Services<br />
Chapter 30. Geriatric Evaluation <strong>and</strong> Management Units for Hospitalized <strong>Patient</strong>s<br />
280
Chapter 26. Prevention <strong>of</strong> Falls in Hospitalized <strong>and</strong> Institutionalized<br />
Older People<br />
Joseph V. Agostini, MD<br />
Dorothy I. Baker, PhD, RNCS<br />
Sidney T. Bogardus, Jr., MD<br />
Yale University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Introduction<br />
A fall is defined as unintentionally coming to rest on the ground, floor, or other lower<br />
level, but not as a result <strong>of</strong> syncope or overwhelming external force. Falling is a common cause<br />
<strong>of</strong> morbidity <strong>and</strong> the leading cause <strong>of</strong> nonfatal injuries <strong>and</strong> trauma-related hospitalizations in the<br />
United States. 1 Complications include bone fractures, injury to the s<strong>of</strong>t tissues, increased<br />
functional dependence, <strong>and</strong> fear <strong>of</strong> falling again, which itself can be debilitating. Each <strong>of</strong> these<br />
complications contributes to increased risk <strong>of</strong> future falls. Studies in community-dwelling older<br />
patients have identified age, gait or balance impairment, sensory or cognitive impairment,<br />
musculoskeletal diseases, environmental hazards, <strong>and</strong> many medications (such as sedativehypnotic<br />
drugs) as risk factors.<br />
One <strong>of</strong> the strongest predictors <strong>of</strong> future falls is having previously fallen. There are<br />
numerous other risk factors for falls in older persons, which are reviewed in detail elsewhere. 2,3<br />
The number <strong>of</strong> risk factors is correlated with the risk <strong>of</strong> falling. A study by Tinetti <strong>and</strong><br />
colleagues found the risk <strong>of</strong> falling increased from 19% when one risk factor was present to 78%<br />
in the presence <strong>of</strong> 4 or more risk factors. 4 Some <strong>of</strong> the factors associated with fall risk in the<br />
hospital setting, however, may differ from those in community-dwelling or institutional settings.<br />
The onset <strong>of</strong> an acute illness leading to hospitalization may increase fall risk due to immobility<br />
<strong>and</strong> deconditioning. Treatment for an acute condition, such as the addition <strong>of</strong> new medications or<br />
an altered medication regimen, may also increase fall risk.<br />
The hospital environment itself may either be a supportive environment (eg, the presence<br />
<strong>of</strong> h<strong>and</strong>rails <strong>and</strong> no-slip bathing surfaces) or may contribute to fall risk (eg, unfamiliar rooms,<br />
improper bed height). This chapter reviews general evidence regarding multicomponent falls<br />
prevention protocols, <strong>and</strong> 5 specific interventions: identification bracelets, physical restraints,<br />
bed alarms, special flooring, <strong>and</strong> hip protectors.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Falls are among the most common incidents reported in institutions, 5 although incident<br />
reports may underestimate their true occurrence. 6 The incidence <strong>of</strong> falls among hospitalized<br />
patients varies according to the risk factors <strong>and</strong> case mix <strong>of</strong> the patient population as well as the<br />
presence <strong>of</strong> falls prevention measures. Rubinstein has reported fall rates <strong>of</strong> 0.6 to 2.9 falls<br />
annually per bed in hospitalized patients <strong>and</strong> 0.6 to 3.6 falls annually per bed in long-term care<br />
institutions, based on published data. 7 About 50% <strong>of</strong> the 1.7 million nursing home residents in<br />
the United States fall at least once each year, resulting in serious injury in about 10% <strong>of</strong><br />
residents. 7-9 The total cost <strong>of</strong> falls injuries in 1994 for adults aged 65 years <strong>and</strong> older was<br />
estimated at $20.2 billion. 10<br />
Hip fractures are the most feared complication <strong>of</strong> falls. Up to 20% <strong>of</strong> people sustaining a<br />
hip fracture become nonambulatory, <strong>and</strong> only 14-21% recover their ability to carry out<br />
281
instrumental activities <strong>of</strong> daily living. 11 The estimated total incremental costs (the difference<br />
between costs before <strong>and</strong> after a hip fracture) <strong>of</strong> caring for an individual in the year after fracture<br />
were estimated to be between $16,300 <strong>and</strong> $18,700. 12 Estimated Medicare expenditures for hip<br />
fractures in 1991 were about $2.9 billion. 13<br />
Practice Description<br />
Based on the multifactorial etiology <strong>of</strong> falls, multicomponent interventions have been<br />
developed to address patient risk factors <strong>and</strong> decrease fall rates. However, most studies have not<br />
been designed in a way to determine which components <strong>of</strong> a multicomponent intervention are<br />
most effective.<br />
Risk Assessment<br />
A variety <strong>of</strong> institution-based programs have been implemented to prevent falls. These<br />
programs usually begin by identifying individuals at increased risk for falling. This is<br />
accomplished by history-taking to elicit past falls history or by using more formal assessment<br />
tools. 4,14-17 Protocols used to perform falls risk assessment in hospitals or nursing homes vary by<br />
institution <strong>and</strong> <strong>of</strong>ten have not been validated. 18<br />
Community-Dwelling Elders<br />
An overwhelming majority <strong>of</strong> the large, prospective, controlled studies have been carried<br />
out in the outpatient environment. They deserve mention because many <strong>of</strong> the interventions<br />
could be modified for a hospital-based intervention. Tinetti <strong>and</strong> colleagues 19 showed that<br />
interventions to reduce specific risk factors resulted in a 30% reduction in falls over one year in a<br />
prospective community cohort. The targeted risk factors were postural hypotension, use <strong>of</strong> any<br />
benzodiazepine or sedative-hypnotic drug, use <strong>of</strong> 4 or more prescription medications,<br />
environmental hazards, <strong>and</strong> muscular strength or range <strong>of</strong> motion impairments. Specific<br />
interventions that were part <strong>of</strong> the multicomponent program included exercise recommendations,<br />
behavioral recommendations, medication review, <strong>and</strong> environmental modifications. A systematic<br />
review <strong>of</strong> predominantly non-hospital based multi-risk factor intervention studies showed<br />
significant protection against falling (Peto OR 0.77, 95% CI: 0.64-0.91). 20 There was, however,<br />
significant heterogeneity across studies.<br />
The large literature on community-based interventions has yielded other insights, some <strong>of</strong><br />
which may be applicable to the acute care setting. For example, exercise-based interventions 21-25<br />
have been studied as a means to decrease falls in older persons. Results <strong>of</strong> these trials have not<br />
been conclusive. A pre-planned meta-analysis <strong>of</strong> 7 r<strong>and</strong>omized controlled trials (2 nursing homebased<br />
<strong>and</strong> 5 community-based) that included an exercise component found a 10% decrease in<br />
fall risk (adjusted incidence ratio 0.90, 95% CI: 0.81-0.99), 26 although a recent systematic review<br />
examining the effect <strong>of</strong> 4 trials <strong>of</strong> exercise alone found no protection against falling. 20 Another<br />
important insight from primarily non-hospital settings includes the association between specific<br />
medications or classes <strong>of</strong> medications <strong>and</strong> falls. 27,28 Although several studies have used<br />
pharmacist- or physician-based medication reviews as part <strong>of</strong> a multifaceted intervention, the<br />
independent effect <strong>of</strong> medication review <strong>and</strong> adjustment on fall outcomes has not been reported.<br />
Institutionalized Elders<br />
In a nursing home setting, a promising r<strong>and</strong>omized controlled trial incorporating<br />
individualized assessment <strong>and</strong> targeting 4 falls-associated domains has been reported. 29<br />
Intervention facilities had 19% fewer recurrent falls (95% CI: 2%-36%) compared with control<br />
282
facilities <strong>and</strong> a 31% reduction in mean rate <strong>of</strong> injurious falls (13.7 vs. 19.9 falls per 100 personyears;<br />
p=0.22). Interventions in this study were made in the areas <strong>of</strong> environmental <strong>and</strong> personal<br />
safety (improvement in room lighting, flooring, footwear), wheelchair use <strong>and</strong> maintenance<br />
(assessment by an occupational therapist), psychotropic drug prescription (assessment <strong>and</strong><br />
recommendations for change), transfer <strong>and</strong> ambulation (evaluation <strong>and</strong> recommendations for<br />
change), <strong>and</strong> facility-wide interventions (eg, in-service educational programs). No analogous<br />
study <strong>of</strong> a multi-intervention st<strong>and</strong>ardized protocol has been reported in hospitalized patients.<br />
Hospitalized Elders<br />
In the hospital, interventions have been employed as part <strong>of</strong> multiple risk factor<br />
intervention studies, but many have been poorly described <strong>and</strong> st<strong>and</strong>ardized. In the studies set in<br />
acute care environments, 30-45 practices include educational activities for nurse <strong>and</strong> support staff,<br />
patient orientation activities, review <strong>of</strong> prior falls, <strong>and</strong> improvement <strong>of</strong> surrounding environment.<br />
Specific environmental components included decreasing ward or room obstacles, adding<br />
supplemental lighting <strong>and</strong> grab bars in bathrooms, <strong>and</strong> lowering bedrails <strong>and</strong> bed height. Other<br />
studies have attempted to improve transfer <strong>and</strong> mobility by providing scheduled ambulatory <strong>and</strong><br />
physical therapy activities <strong>and</strong> better footwear (eg, non-skid socks). Additionally, studies have<br />
incorporated strategies to assist cognitively impaired patients by educating family members to<br />
deal with confused patients, minimizing sedating medications, <strong>and</strong> moving confused patients<br />
closer to nursing staff. Because many <strong>of</strong> these hospital studies use small sample sizes <strong>and</strong><br />
inadequately describe the precise number <strong>and</strong> st<strong>and</strong>ardization <strong>of</strong> interventions, their<br />
generalizability <strong>and</strong> reproducibility is limited. However, a recent systematic review <strong>of</strong> many <strong>of</strong><br />
these programs concluded that a pooled effect <strong>of</strong> 25% reduction in the fall rate occurred in the<br />
studies that examined prospective interventions compared to fall risk in historical controls. 18<br />
Some interventions with the potential for effectiveness in isolation have been studied.<br />
Each <strong>of</strong> the following hospital- or institution-based individual interventions has been analyzed<br />
independently <strong>of</strong> a multi-component falls prevention program:<br />
• Identification bracelets<br />
• Physical restraints<br />
• Bed alarms<br />
• Special flooring<br />
• Hip protectors<br />
Several generally accepted interventions with high face-validity have not been<br />
independently studied, yet are commonly accepted practices. Immobility 46 is a significant risk<br />
factor for several geriatric complications, including falls, pressure ulcers, <strong>and</strong> functional decline.<br />
Minimization <strong>of</strong> bedrest is a practical, real-world intervention that has implications for<br />
prevention <strong>of</strong> a number <strong>of</strong> serious hospital-acquired complications. 47<br />
Comment<br />
There are few hospital or other institution-based r<strong>and</strong>omized controlled trials <strong>of</strong><br />
st<strong>and</strong>ardized falls interventions, although the necessity for well-designed studies is clear. The<br />
nursing home-based intervention reported by Ray <strong>and</strong> colleagues 29 provides good evidence that a<br />
well-documented intervention can improve falls outcomes in institutionalized patients. No<br />
similarly designed trial <strong>of</strong> a multicomponent intervention in hospitalized patients was identified,<br />
although many falls prevention programs incorporate multifactorial interventions. The questions<br />
raised by multicomponent falls prevention studies include the generalizability <strong>of</strong> interventions to<br />
283
diverse inpatient settings, appropriate targeting <strong>of</strong> at-risk individuals, analysis <strong>of</strong> the individual<br />
components that provide the best improvement in falls outcomes, <strong>and</strong> the transportability <strong>of</strong><br />
interventions between institutions with variable resources for implementation. Evidence for the<br />
effectiveness <strong>of</strong> individual interventions is important, but effectiveness may change (for better or<br />
worse) when such interventions are incorporated with others as part <strong>of</strong> a falls prevention<br />
program.<br />
Subchapter 26.1. Identification Bracelets for High-Risk <strong>Patient</strong>s<br />
Background<br />
Some hospitals use colored bracelets to identify patients at high risk for falls. Other<br />
identification methods include signs, stickers, or tags placed above the patient’s bed, at the<br />
nursing station, or on the patient’s chart. In theory, these remind staff that the patient is at high<br />
risk for falls <strong>and</strong> trigger interventions that reduce the risk <strong>of</strong> falls (eg, supervision or assistance<br />
with ambulation, minimization <strong>of</strong> sedative-hypnotic medications, lowering <strong>of</strong> bed height).<br />
Identification bracelets might also impact patients’ falls awareness (eg, reminding patients to call<br />
for assistance before getting out <strong>of</strong> bed).<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
See Introduction to Chapter 26.<br />
Opportunities for Impact<br />
We found no published data on the number <strong>of</strong> hospitals currently using such strategies.<br />
Practice Description <strong>and</strong> Evidence for Effectiveness<br />
A search <strong>of</strong> the literature identified many studies that have used identification bracelets,<br />
signs, or tags for high-risk patients. 31-33,35,40-42,44,45,48,49 Most <strong>of</strong> these involved multiple,<br />
simultaneous interventions <strong>and</strong> were designed such that estimation <strong>of</strong> the treatment effect due to<br />
the identification bracelet, signs or tags component cannot be calculated. The remaining study<br />
was a r<strong>and</strong>omized, controlled trial <strong>of</strong> colored identification bracelets worn by inpatients at high<br />
risk for falls (Table 26.1.1). 50 “High-risk” was defined as history <strong>of</strong> multiple falls, an episode <strong>of</strong><br />
incontinence, or an admitting diagnosis <strong>of</strong> stroke or ataxia. Cox proportional hazards model was<br />
used to assess the effect <strong>of</strong> identification bracelets on time-to-first-fall. The fall rate was 42%<br />
(27/65) in the intervention group <strong>and</strong> 30% (21/69) in the control group, which did not represent a<br />
statistically significant difference. After preliminary analysis <strong>of</strong> the data, the investigators <strong>and</strong><br />
ethics committee agreed that it was not appropriate to continue for the sole purpose <strong>of</strong> obtaining<br />
statistical power, <strong>and</strong> the study was terminated.<br />
Potential for Harm<br />
None identified.<br />
Costs <strong>and</strong> Implementation<br />
Identification tags <strong>and</strong> similar interventions are associated with minimal costs.<br />
Comment<br />
Use <strong>of</strong> special bracelets, signs, <strong>and</strong> stickers to identify patients at high risk for falls is a<br />
relatively inexpensive <strong>and</strong> easy to implement practice. There is currently insufficient information<br />
284
as to whether identification bracelets, as a isolated intervention, decrease falls. Future studies<br />
should assess the effectiveness <strong>of</strong> similar identification strategies in the context <strong>of</strong><br />
multicomponent fall prevention programs <strong>and</strong>, if they are effective, which methods work best.<br />
Table 26.1.1. Study <strong>of</strong> identification bracelets*<br />
Study Participants <strong>and</strong> Setting Study Design,<br />
Outcomes<br />
Mayo,<br />
1994 50<br />
134 high-risk patients in a<br />
rehabilitation hospital, 1990-91<br />
* CI indicates confidence interval.<br />
References<br />
Level 1,<br />
Level 1<br />
285<br />
Results<br />
Hazard ratio for fall with<br />
intervention: 1.3 (95% CI: 0.8-2.4)<br />
1. Baker SP, O'Neill B, Ginsburg MJ, Guohua L. The injury fact book. 2nd edition. New<br />
York: Oxford University Press, 1992.<br />
2. Tinetti ME, Speechley M. Prevention <strong>of</strong> falls among the elderly. N Engl J Med.<br />
1989;320:1055-9.<br />
3. Myers AH, Young Y, Langlois JA. Prevention <strong>of</strong> falls in the elderly. Bone. 1996;18:87S-<br />
101S.<br />
4. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living<br />
in the community. N Engl J Med. 1988;319:1701-1707.<br />
5. Sutton J, St<strong>and</strong>an P, Wallace A. Incidence <strong>and</strong> documentation <strong>of</strong> patient accidents in<br />
hospital. Nurs Times. 1994;90:29-35.<br />
6. Kanten DN, Mulrow CD, Gerety MB, Lichtenstein MJ, Aguilar C, Cornell JE. Falls: an<br />
examination <strong>of</strong> three reporting methods in nursing homes. J Am Geriatr Soc.<br />
1993;41:662-666.<br />
7. Rubenstein LZ, Robbins AS, Schulman BL, Rosado J, Osterweil D, Josephson KR. Falls<br />
<strong>and</strong> instability in the elderly. J Am Geriatr Soc.1988;36:266-278.<br />
8. Rubenstein LZ, Josephson KR, Robbins AS. Falls in the nursing home. Ann Intern Med.<br />
1994;121:442-451.<br />
9. Tinetti ME. Factors associated with serious injury during falls by ambulatory nursing<br />
home residents. J Am Geriatr Soc. 1987;35:644-648.<br />
10. Engl<strong>and</strong>er F, Hodson TJ, Terregrossa RA. Economic dimensions <strong>of</strong> slip <strong>and</strong> fall injuries.<br />
J Forensic Sci. 1996; 41:733-746.<br />
11. Zuckerman JD. Hip fracture. N Engl J Med. 1996;334:1519-1525.<br />
12. Brainsky GA, Lydick E, Epstein R, et al. The economic cost <strong>of</strong> hip fractures in<br />
community-dwelling older adults: a prospective study. J Am Geriatr Soc.1997;45:281-<br />
287.<br />
13. CDC. Incidence <strong>and</strong> costs to Medicare <strong>of</strong> fractures among Medicare beneficiaries aged ><br />
65 years: United States, July 1991-June 1992. MMWR - Morbidity & Mortality Weekly<br />
Report 1996;45:877-883.<br />
14. Oliver D, Britton M, Seed P, Martin FC, Hopper AH. Development <strong>and</strong> evaluation <strong>of</strong><br />
evidence based risk assessment tool (STRATIFY) to predict which elderly inpatients will<br />
fall: case-control <strong>and</strong> cohort studies. BMJ. 1997;315:1049-1053.<br />
15. Whitney SL, Poole JL, Cass SP. A review <strong>of</strong> balance instruments for older adults. Am J<br />
Occup Ther. 1998;52: 666-671.
16. Shumway-Cook A, Baldwin M, Polissar NL, Gruber W. Predicting the probability for<br />
falls in community-dwelling older adults. Phys Ther. 1997;77:812-819.<br />
17. Tinetti ME. Performance-oriented assessment <strong>of</strong> mobility problems in elderly patients. J<br />
Am Geriatr Soc. 1986;34:119-126.<br />
18. Oliver D, Hopper A, Seed P. Do hospital fall prevention programs work? A systematic<br />
review. J Am Geriatr Soc. 2000;48:1679-1689.<br />
19. Tinetti ME, Baker DI, McAvay G, et al. A multifactorial intervention to reduce the risk<br />
<strong>of</strong> falling among elderly people living in the community. N Engl J Med. 1994;331:821-<br />
827.<br />
20. Gillespie LD, Gillespie WJ, Cumming R, Lamb SE, Rowe BH. Interventions for<br />
preventing falls in the elderly. In: The Cochrane Library, Issue 2, 2000. Oxford: Update<br />
S<strong>of</strong>tware.<br />
21. McMurdo ME, Mole P, Paterson C. Controlled trial <strong>of</strong> weight-bearing exercise in older<br />
women in relation to bone density <strong>and</strong> falls. BMJ. 1997;314:569.<br />
22. Wolf SL, Barnhart HX, Kutner NG, McNeely E, Coogler C, Xu T. Reducing frailty <strong>and</strong><br />
falls in older persons: an investigation <strong>of</strong> Tai Chi <strong>and</strong> computerized balance training.<br />
Atlanta FICSIT Group. Frailty <strong>and</strong> Injuries: Cooperative Studies <strong>of</strong> Intervention<br />
Techniques. J Am Geriatr Soc.1996;44:489-497.<br />
23. Lord SR, Ward JA, Williams P, Strudwick M. The effect <strong>of</strong> a 12-month exercise trial on<br />
balance, strength, <strong>and</strong> falls in older women: a r<strong>and</strong>omized controlled trial. J Am Geriatr<br />
Soc. 1995;43:1198-1206.<br />
24. Reinsch S, MacRae P, Lachenbruch PA, Tobis JS. Attempts to prevent falls <strong>and</strong> injury: a<br />
prospective community study. Gerontologist. 1992;32:450-456.<br />
25. Mulrow CD, Gerety MB, Kanten D, et al. A r<strong>and</strong>omized trial <strong>of</strong> physical rehabilitation<br />
for very frail nursing home residents. JAMA. 1994;271:519-524.<br />
26. Province MA, Hadley EC, Hornbrook MC, et al. The effects <strong>of</strong> exercise on falls in<br />
elderly patients. A preplanned meta-analysis <strong>of</strong> the FICSIT Trials. Frailty <strong>and</strong> Injuries:<br />
Cooperative Studies <strong>of</strong> Intervention Techniques. JAMA. 1995;273:1341-1347.<br />
27. Leipzig RM, Cumming RG, Tinetti ME. Drugs <strong>and</strong> falls in older people: a systematic<br />
review <strong>and</strong> meta-analysis: I. Psychotropic drugs. J Am Geriatr Soc.1999;47:30-39.<br />
28. Leipzig RM, Cumming RG, Tinetti ME. Drugs <strong>and</strong> falls in older people: a systematic<br />
review <strong>and</strong> meta-analysis: II. Cardiac <strong>and</strong> analgesic drugs. J Am Geriatr Soc. 1999;47:40-<br />
50.<br />
29. Ray WA, Taylor JA, Meador KG, et al. A r<strong>and</strong>omized trial <strong>of</strong> a consultation service to<br />
reduce falls in nursing homes. JAMA. 1997;278:557-562.<br />
30. Rainville NG. Effect <strong>of</strong> an implemented fall prevention program on the frequency <strong>of</strong><br />
patient falls. Qual Rev Bull. 1984;10:287-291.<br />
31. Fife DD, Solomon P, Stanton M. A risk/falls program: code orange for success...geriatric<br />
patients. Nurs Manage. 1984;15:50-53.<br />
32. Barker SM, O'Brien CN, Carey D, Weisman GK. Quality improvement in action: a falls<br />
prevention <strong>and</strong> management program. Mt Sinai J Med. 1993;60:387-90.<br />
33. Hendrich AL. An effective unit-based fall prevention plan. J Nurs Qual Assur.<br />
1988;3:28-36.<br />
34. Craighead J, Fletcher P, Maxwell J. Seven steps for fall prevention. Dimens Health Serv.<br />
1991;68:25-26.<br />
35. Kilpack V, Boehm J, Smith N, Mudge B. Using research-based nursing interventions to<br />
decrease patient falls. Appl Nurs Res. 1991;4:50-56.<br />
286
36. Mitchell A, Jones N. Striving to prevent falls in an acute care setting–action to enhance<br />
quality. J Clin Nurs. 1996;5:213-220.<br />
37. Hill BA, Johnson R, Garrett BJ. Reducing the incidence <strong>of</strong> falls in high risk patients. J<br />
Nurs Adm. 1988;18:24-28.<br />
38. Hern<strong>and</strong>ez M, Miller J. How to reduce falls. Geriatr Nurs. 1986;7:97-102.<br />
39. Krishna KM, Van Cleave RJ. Decrease in the incidence <strong>of</strong> patient falls in a geriatric<br />
hospital after educational programs [letter]. J Am Geriatr Soc.1983;31:187.<br />
40. Morton D. Five years <strong>of</strong> fewer falls. Am J Nurs. 1989;89:204-205.<br />
41. Sweeting HL. <strong>Patient</strong> fall prevention–a structured approach. J Nurs Manag. 1994;2:187-<br />
92.<br />
42. Schmid NA. Reducing patient falls: a research-based comprehensive fall prevention<br />
program. Mil Med. 1990;155:202-207.<br />
43. Brady R, Chester FR, Pierce LL, Salter JP, Schreck S, Radziewicz R. Geriatric falls:<br />
prevention strategies for the staff. J Gerontol Nurs. 1993;19:26-32.<br />
44. Cannard G. Falling trend. Nurs Times. 1996;92:36-37.<br />
45. Zepp S. "Ban a fall." A nursing innovation to reducing patient falls. Kans Nurse.<br />
1991;66:13.<br />
46. Mahoney JE. Immobility <strong>and</strong> falls. Clin Geriatr Med.1998;14:699-726.<br />
47. Harper CM, Lyles YM. Physiology <strong>and</strong> complications <strong>of</strong> bed rest. J Am Geriatr Soc.<br />
1988;36:1047-1054.<br />
48. Rogers S. Reducing falls in a rehabilitation setting: a safer environment through team<br />
effort. Rehabil Nurs. 1994;19:274-276,322.<br />
49. Innes E. Maintaining fall prevention. QRB. 1985;11:217-221.<br />
50. Mayo NE, Gloutney L, Levy AR. A r<strong>and</strong>omized trial <strong>of</strong> identification bracelets to<br />
prevent falls among patients in a rehabilitation hospital. Arch Phys Med Rehabil.<br />
1994;75:1302-1308.<br />
Subchapter 26.2. Interventions that Decrease the Use <strong>of</strong> Physical Restraints<br />
Background<br />
The Health Care Financing Administration (HCFA) defines physical restraints as “any<br />
manual method or physical or mechanical device, material, or equipment attached or adjacent to<br />
the patient that the individual cannot remove easily which restricts freedom <strong>of</strong> movement or<br />
normal access to one’s body.” 1 Physical restraints have been used in nursing homes <strong>and</strong> hospitals<br />
both as a safety device <strong>and</strong> as a falls prevention tool. Because restrained patients cannot arise<br />
from a chair or transfer out <strong>of</strong> bed, they theoretically will not fall or, in the case <strong>of</strong> bedrails, will<br />
not roll out <strong>of</strong> bed. However, the use <strong>of</strong> physical restraints may lead to substantial adverse<br />
outcomes. In fact, serious injuries <strong>and</strong> even death have been reported with use <strong>of</strong> these devices. 2,3<br />
This chapter examines interventions to reduce use <strong>of</strong> physical restraints <strong>and</strong> the concomitant<br />
effect on fall rates.<br />
Practice Description<br />
Studies examining the use physical restraints have considered 2 types <strong>of</strong> interventions in<br />
hospital or nursing home settings: bedrails <strong>and</strong> other mechanical restraints designed to restrict<br />
mobility. These interventions usually begin with either a physician or nurse making an<br />
assessment that a patient is at risk for falls, elopement, or other adverse outcomes. Thereafter,<br />
287
use <strong>of</strong> a restraint is initiated, with periodic reassessment <strong>of</strong> the ongoing need for the device.<br />
<strong>Safety</strong> practices to reduce restraint use in nursing home patients have included nursing education<br />
strategies focusing on assessment/reassessment <strong>of</strong> the need for restraints <strong>and</strong> the use <strong>of</strong><br />
alternatives to restraints.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
See Introduction to Chapter 26.<br />
Opportunities for Impact<br />
Federal guidelines now discourage all but the limited, appropriate use <strong>of</strong> physical<br />
restraints <strong>and</strong> bedrails. Legislation adopted as part <strong>of</strong> the Omnibus Budget Reconciliation Act <strong>of</strong><br />
1987 directed nursing homes to limit physical restraints, <strong>and</strong> the Joint Commission on the<br />
Accreditation <strong>of</strong> Healthcare Organizations (JCAHO) has adopted similar guidelines. Several<br />
statewide initiatives (eg, the Pennsylvania Restraint Reduction Initiative, begun in 1996) have<br />
been implemented under HCFA’s National Restraint Reduction Initiative, resulting in notable<br />
reductions in restraint usage. 4 The Food <strong>and</strong> Drug Administration’s Hospital Bed <strong>Safety</strong> Work<br />
Group has likewise actively raised awareness <strong>of</strong> the risks <strong>and</strong> benefits <strong>of</strong> bedrail use. 5 Based on<br />
an annual HCFA survey, the national restraint rate was approximately 13.5% in 1999, down<br />
from approximately 20% in 1996 when HCFA’s Restraint Reduction Initiative began. 6<br />
Nonetheless, data from selected states reveals that the rate was still as high as 26% as <strong>of</strong> 1998. 7<br />
Study Designs <strong>and</strong> Outcomes<br />
Six studies were identified: 2 concerning bedrail interventions 8,9 <strong>and</strong> 4 describing<br />
mechanical restraints interventions (Table 26.2.1). 7,10-12 Most studies compare interventions with<br />
historical control or baseline rates using a before-after study design. Morbidity data on falls are<br />
reported in all studies.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The studies reveal no statistically significant difference in falls compared with historical<br />
controls when bedrails are removed. In fact, restrained patients appear to have a modest increase<br />
in fall risk or fall injuries based on several studies. Weaknesses in study design for some <strong>of</strong> these<br />
studies preclude a final conclusion.<br />
Potential for Harm<br />
The potential for harm with use <strong>of</strong> bedrails is well-documented, including death from a<br />
variety <strong>of</strong> mechanisms, including death <strong>and</strong> strangulation. 13 Mechanical restraints likewise carry<br />
a risk <strong>of</strong> severe injury, strangulation, <strong>and</strong> mobility limitations that may predispose patients to<br />
other adverse outcomes (pressure ulcers, incontinence, acute confusion). Limits to patient<br />
freedom, dignity, <strong>and</strong> quality <strong>of</strong> life also contribute to the potential for harm. A potential harm <strong>of</strong><br />
interventions to decrease restraint use is that there may be an increase in other adverse events<br />
(eg, elopement) if appropriate alternative preventive measures are not in place.<br />
Costs <strong>and</strong> Implementation<br />
The costs associated with interventions to reduce the use <strong>of</strong> restraints have not been<br />
described. Nonetheless, reduction in the use <strong>of</strong> physical restraints will require resources to pay<br />
for alternative interventions <strong>and</strong> rehabilitative measures <strong>and</strong> will increase labor costs. 14<br />
Compliance with interventions to reduce bedrail rates <strong>and</strong> to decrease mechanical restraint use<br />
288
has been good. In fact, given adequate alternatives to the use <strong>of</strong> these devices, hospital <strong>and</strong><br />
nursing staffs have decreased their usage significantly. In the Neufeld study, 7 for example,<br />
restraint use fell from 41% to 4%.<br />
Comment<br />
There is growing evidence that physical restraints have a limited role in medical care.<br />
Restraints limit mobility, a shared risk factor for a number <strong>of</strong> adverse geriatric outcomes, <strong>and</strong><br />
increase the risk <strong>of</strong> iatrogenic events. They certainly do not eliminate falls, <strong>and</strong> decreasing their<br />
use can be accomplished without increasing fall rates. In some instances reducing the use <strong>of</strong><br />
restraints may actually decrease the risk <strong>of</strong> falling. Incorporating changes into physician <strong>and</strong> staff<br />
behavior may be easier if large, multicenter trials are successful in identifying safe alternatives to<br />
restraints that effectively limit falls risks for patients.<br />
289
Table 26.2.1. Studies <strong>of</strong> physical restraints <strong>and</strong> fall risk*<br />
Study Participants <strong>and</strong> Setting Design,<br />
Outcomes<br />
Hanger,<br />
1999 9<br />
1968 hospital patients in New<br />
Zeal<strong>and</strong>, 1994; formal bedrail<br />
policy <strong>and</strong> educational<br />
program to reduce bedrail use,<br />
historical controls<br />
Si, 1999 8 246 patients in a teaching<br />
nursing home, 1993-94;<br />
interdisciplinary team<br />
assessment <strong>and</strong> removal <strong>of</strong><br />
bedrails with provision <strong>of</strong><br />
bedrail alternatives, historical<br />
controls<br />
Capezuti,<br />
1996 11<br />
Capezuti,<br />
1998 12<br />
Neufeld,<br />
1999 7<br />
Tinetti,<br />
1991 10<br />
322 nursing home residents;<br />
subgroup <strong>of</strong> confused patients<br />
examined for mechanical<br />
restraint use<br />
633 nursing home residents in<br />
3 nursing homes, 1990-1991;<br />
restraint education <strong>and</strong><br />
consultation interventions<br />
compared with baseline rates<br />
2075 nursing home beds in 16<br />
nursing homes, 1991-1993;<br />
educational intervention to<br />
decrease mechanical restraints<br />
compared with baseline rates<br />
397 elderly patients at 12<br />
skilled nursing facilities;<br />
observational cohort study <strong>of</strong><br />
mechanical restraint use<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
* CI indicates confidence interval; OR, odds ratio.<br />
290<br />
Results<br />
No significant difference in overall fall<br />
rate: 164.8 falls/10,000 bed days<br />
before <strong>and</strong> 191.7 falls/10,000 bed<br />
days after the intervention (p=0.18)<br />
Fewer serious falls occurred after the<br />
intervention (p=0.008)<br />
No significant difference in fall rates:<br />
2/116 (1.7%) patients before <strong>and</strong><br />
2/130 (1.5%) patients after the<br />
intervention<br />
Confused patients who were restrained<br />
had increased odds <strong>of</strong> falling (OR<br />
1.65, 95% CI: 0.69-3.98) <strong>and</strong><br />
recurrent falls (OR 2.46, 95% CI:<br />
1.03-5.88)<br />
No significant increase in fall rates in<br />
the restraint-free group<br />
Decreased odds <strong>of</strong> minor injury after<br />
restraint removal, adjusted OR 0.3<br />
(95% CI: 0.1-0.9)<br />
Moderate/severe injuries decreased<br />
from 7.4% to 4.4% (p=0.0004) after<br />
educational intervention<br />
15/275 (5%) <strong>of</strong> unrestrained patients<br />
compared to 21/122 (17%)<br />
experienced a serious fall-related<br />
injury (p
References<br />
1. Health Care Financing Administration. FY 2001 Annual Performance Plan. Available at:<br />
http://www.hcfa.gov/stats/2001.htm. Accessed June 13, 2001.<br />
2. FDA Center for Devices <strong>and</strong> Radiological Health. FDA safety alert: entrapment hazards<br />
with hospital bed side rails. August 1995. Available at: http://www.fda.gov/cdrh/<br />
bedrails.html. Accessed June 12, 2001.<br />
3. Miles SH, Irvine P. Deaths caused by physical restraints. Gerontologist. 1992;32:762-766.<br />
4. Health Care Financing Administration. National restraint reduction newsletter. Summer<br />
2000. Available at: http://www.hcfa.gov/publications/newsletters/restraint/2000/<br />
rr0600.asp. Accessed June 12, 2001.<br />
5. FDA Center for Devices <strong>and</strong> Radiological Health. A guide to bed safety. Available at:<br />
http://www.fda.gov/cdrh/beds/bedrail.pdf. Accessed June 12, 2001.<br />
6. Health Care Financing Administration. National restraint reduction newsletter. Winter<br />
1999. Available at: http://www.hcfa.gov/publications/newsletters/restraint/1999/<br />
rrwin99.htm. Accessed June 12, 2001.<br />
7. Neufeld RR, Libow LS, Foley WJ, Dunbar JM, Cohen C, Breuer B. Restraint reduction<br />
reduces serious injuries among nursing home residents. J Am Geriatr Soc. 1999;47:1202-<br />
1207.<br />
8. Si M, Neufeld RR, Dunbar J. Removal <strong>of</strong> bedrails on a short-term nursing home<br />
rehabilitation unit. Gerontologist. 1999;39:611-614.<br />
9. Hanger HC, Ball MC, Wood LA. An analysis <strong>of</strong> falls in the hospital: can we do without<br />
bedrails? J Am Geriatr Soc. 1999;47:529-531.<br />
10. Tinetti ME, Liu YB, Ginter S. Mechanical restraint use <strong>and</strong> fall related injuries among<br />
residents <strong>of</strong> skilled nursing facilities. Ann Intern Med. 1992;116:369-374.<br />
11. Capezuti E, Evans L, Strumpf N, Maislin G. Physical restraint use <strong>and</strong> falls in nursing<br />
home residents. J Am Geriatr Soc. 1996;44:627-633.<br />
12. Capezuti E, Strumpf NE, Evans LK, Grisso JA, Maislin G. The relationship between<br />
physical restraint removal <strong>and</strong> falls <strong>and</strong> injuries among nursing home residents. J Gerontol<br />
A Biol Sci Med Sci. 1998;53:M47-M52.<br />
13. Parker K, Miles SH. Deaths caused by bedrails. J Am Geriatr Soc. 1997;45:797-802.<br />
14. Schnelle JF, Smith RL. To use physical restraints or not? [editorial]. J Am Geriatr Soc.<br />
1996;44:727-728.<br />
Subchapter 26.3. Bed Alarms<br />
Background<br />
Epidemiologic studies reveal that falls occur commonly in <strong>and</strong> around bed areas. 1,2<br />
Decreasing the risk <strong>of</strong> falls when patients attempt to transfer into <strong>and</strong> out <strong>of</strong> bed without<br />
assistance is a potentially important target safety goal. This chapter examines the use <strong>of</strong> a bed<br />
alarm system that alerts hospital staff to patient movement out <strong>of</strong> bed as a strategy to reduce<br />
falls. General principles <strong>of</strong> alarm use in health care settings can be found in Chapter 8.<br />
291
Practice Description<br />
A sensor device is placed on the bed, under a sitting or reclining patient. When the<br />
patient changes position, it detects movement <strong>and</strong>/or absence <strong>of</strong> weight. An audible alarm is<br />
triggered at the nurses’ station <strong>and</strong>, with some devices, in the patient’s room. The alarm alerts<br />
nurses when patients attempt to leave the bed without assistance <strong>and</strong> may alert a patient to<br />
remain in bed if the alarm is audible in the patient’s room.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Several studies have included bed alarms as part <strong>of</strong> a multifaceted intervention. 3-6<br />
However, the study designs do not allow calculation <strong>of</strong> the effect attributable to the bed alarm<br />
component or were not controlled. A recent, unpublished before-after study was identified in a<br />
Web search but the full report could not be obtained before completion <strong>of</strong> this chapter. 7<br />
Tideiksaar et al r<strong>and</strong>omized elderly patients at “high risk” for falls to either a group that received<br />
an alarm system (the RN+ OnCall bed monitoring system) or to a control group that did not<br />
(Table 26.3.1) 8 . The groups were similar in age <strong>and</strong> gender. No other baseline comparisons were<br />
reported. There were fewer falls in the study group but the difference failed to reach statistical<br />
significance. However, the total number <strong>of</strong> falls was low (n=17) <strong>and</strong> had there been one less fall<br />
in the alarm group or one more fall in the control group, the difference would have been<br />
statistically significant.<br />
Potential for Harm<br />
No harm was identified. There are theoretical electrical risks if the sensor devices are<br />
internally compromised due to bending <strong>of</strong> the sensor mats <strong>and</strong> exposure to fluids, but such<br />
events have not been reported in the literature.<br />
Costs <strong>and</strong> Implementation<br />
Costs <strong>of</strong> the devices vary by manufacturer, the type <strong>of</strong> bed monitoring system used, <strong>and</strong><br />
the number <strong>of</strong> beds to be monitored. Manufacturers’ charges range from several hundred to<br />
several thous<strong>and</strong> dollars for the receiving equipment. Individual sensors require replacement<br />
after pre-specified periods <strong>of</strong> use or, in some cases, can be cleaned between patients, which<br />
incurs additional hospital costs. Implementation requires adequate staffing to respond in a timely<br />
manner to the audible alarms.<br />
Comment<br />
At this time, there is insufficient evidence regarding the effectiveness <strong>of</strong> bed alarms in<br />
preventing falls in elderly patients to recommend the practice. Additional research sufficiently<br />
powered to identify meaningful differences, coupled with a formal economic analysis, would be<br />
useful.<br />
292
Table 26.3.1. Study <strong>of</strong> bed alarms for fall prevention<br />
Study Participants <strong>and</strong> Setting Study Design,<br />
Outcomes<br />
Tideiksaar,<br />
1993 8<br />
70 patients on a geriatric unit in<br />
a university hospital, 1992<br />
Level 1,<br />
Level 1<br />
293<br />
Results<br />
(95% Confidence Interval)<br />
Odds ratio for prevention <strong>of</strong><br />
falls: 0.32 (0.10-1.03)<br />
References<br />
1. Vassallo M, Amersey RA, Sharma JC, Allen SC. Falls on integrated medical wards.<br />
Gerontology. 2000;46:158-162.<br />
2. Sutton J, St<strong>and</strong>an P, Wallace A. Incidence <strong>and</strong> documentation <strong>of</strong> patient accidents in<br />
hospital. Nurs Times. 1994;90:29-35.<br />
3. Schmid NA. Reducing patient falls: a research-based comprehensive fall prevention<br />
program. Mil Med. 1990;155:202-207.<br />
4. Widder B. A new device to decrease falls. Geriatr Nurs. 1985;6:287-288.<br />
5. Morton D. Five years <strong>of</strong> fewer falls. Am J Nurs. 1989;89:204-205.<br />
6. Innes E. Maintaining fall prevention. QRB. 1985;11:217-221.<br />
7. Bed-Check Corporation. Fall prevention study: bed alarms, investigating their impact on<br />
fall prevention <strong>and</strong> restraint use. Available at: http://www.bedcheck.com/fall-preventionstudy.html.<br />
Accessed June 6, 2001.<br />
8. Tideiksaar R, Feiner CF, Maby J. Falls prevention: the efficacy <strong>of</strong> a bed alarm system in an<br />
acute-care setting. Mt Sinai J Med. 1993;60:522-527.<br />
Subchapter 26.4. Special Hospital Flooring Materials to Reduce Injuries from <strong>Patient</strong><br />
Falls<br />
Background<br />
One proposed practice to prevent injury due to falls is to alter flooring material on<br />
hospital wards or in nursing homes. Carpeting, vinyl, or other biomedically-engineered materials<br />
could potentially improve falls outcomes. The use <strong>of</strong> special flooring materials has been shown<br />
to influence specific gait characteristics in hospitalized elders. 1 A recent study described the<br />
Penn State <strong>Safety</strong> Floor, which is designed to remain relatively rigid under normal walking<br />
conditions but to deform elastically to absorb impact forces during a fall. 2 The efficacy <strong>of</strong> this<br />
floor is still being tested outside the laboratory environment among nursing home residents. 3<br />
Practice Description<br />
As data on the efficiency <strong>of</strong> the Penn State <strong>Safety</strong> Floor 2 are not yet available, we restrict<br />
our review to the use <strong>of</strong> hospital-duty carpeting compared with “usual” vinyl flooring.
Study Designs <strong>and</strong> Outcomes<br />
We identified 2 studies <strong>of</strong> the effect <strong>of</strong> flooring type (carpet vs. “usual” vinyl flooring) on<br />
falls: a r<strong>and</strong>omized controlled trial in an inpatient rehabilitation unit 4 <strong>and</strong> a retrospective study <strong>of</strong><br />
accidents reported in a care <strong>of</strong> the elderly unit (Table 26.4.1). 5 Both studies reported Level 1<br />
outcomes. The r<strong>and</strong>omized trial measured the rate <strong>of</strong> falls. The retrospective analysis studied<br />
fall-related injury, defined as any graze, bruise, laceration, fracture or pain.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The r<strong>and</strong>omized trial by Donald et al found more falls in the group housed in rooms with<br />
carpeted flooring, although the difference barely failed to achieve statistical significance. The<br />
earlier retrospective analysis by Healey found that the rate <strong>of</strong> injury was significantly lower for<br />
patients who fell on carpet rather than vinyl flooring. 5 The severity <strong>of</strong> injuries was not reported<br />
<strong>and</strong> it was not possible to determine whether the rate <strong>of</strong> falls differed according to flooring<br />
material.<br />
Potential for Harm<br />
No harm was identified, although it is possible that asthmatic patients might react to<br />
increased levels <strong>of</strong> dust-mite allergens in carpeted wards. 6<br />
Costs <strong>and</strong> Implementation<br />
No cost estimates for changes in flooring were reported in the literature. Implementation<br />
<strong>of</strong> this practice would require a large expenditure for facilities upgrades nationwide. Likewise,<br />
the costs associated with keeping various floor surfaces clean in the hospital or nursing home<br />
environment would also be high.<br />
Comment<br />
Advances in biomedical engineering could result in potentially significant redesign <strong>of</strong> the<br />
physical environment in hospitals <strong>and</strong> nursing facilities. The primary aim <strong>of</strong> specialized flooring<br />
could be either to reduce the risk <strong>of</strong> falling or to reduce the risk <strong>of</strong> an injury once a fall has<br />
occurred, or both. The two studies analyzed seem to indicate that carpeted floors may increase<br />
fall rates but decrease fall injuries; it is possible that other surfaces would yield better results.<br />
Further study <strong>of</strong> this area is warranted.<br />
294
Table 26.4. Study <strong>of</strong> special flooring for falls prevention<br />
Study Participants <strong>and</strong> Setting Study Design,<br />
Outcomes<br />
Donald,<br />
2000 4<br />
Healey,<br />
1994 5<br />
32 patients in an elderly care<br />
rehabilitation ward in the<br />
United Kingdom in 1996<br />
R<strong>and</strong>om sample <strong>of</strong> accident<br />
forms (n=213) from care <strong>of</strong><br />
elderly unit over 4 years<br />
Level 2,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
295<br />
Results<br />
Rate <strong>of</strong> falls:<br />
Carpet, 10/16 (63%); vinyl, 1/16 (6%)<br />
RR 8.3 (95% CI: 0.95-73; p=0.05)<br />
Falls resulting in injury:<br />
Carpet, 15%; vinyl, 91% (p
Practice Description<br />
External hip protectors are usually made with plastic pads or shields that are padded or<br />
constructed with foam-type materials. They fit into specially-designed pockets in undergarments<br />
or pants. They are designed to be worn during the day for people who are out <strong>of</strong> bed, walking or<br />
engaged in activities that place them at higher risk for falls. Ideally, they would be worn all the<br />
time to protect individuals from nighttime falls.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
See Introduction to Chapter 26.<br />
Opportunities for Impact<br />
No data on the nationwide use <strong>of</strong> hip protectors in the hospital or nursing home are<br />
available. A small minority <strong>of</strong> institutions are in the process <strong>of</strong> evaluating them, <strong>and</strong> a few may<br />
have begun to use them.<br />
Study Designs<br />
Five relevant r<strong>and</strong>omized controlled trials 4-8 were identified from a literature search <strong>and</strong><br />
from a Cochrane systematic review. 9 The Cochrane review cites 2 additional abstracts 10,11 not<br />
included here. Four <strong>of</strong> the trials evaluate effectiveness <strong>of</strong> the devices <strong>and</strong> one study 8 examines<br />
compliance rates <strong>of</strong> wearing hip protectors as part <strong>of</strong> a pilot study. Two studies were clusterr<strong>and</strong>omized<br />
<strong>and</strong> 2 were r<strong>and</strong>omized by individual patient.<br />
Study Outcomes<br />
Studies reported hip fractures as an outcome, although compliance with the intervention<br />
was the primary outcome in one study. Additional outcomes reported were mortality, falls, <strong>and</strong><br />
non-hip fractures.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
External hip protectors appear to be an effective means to reduce the risk <strong>of</strong> a hip fracture<br />
in older persons aged 65 <strong>and</strong> over who fall. Table 26.5.1 lists the abstracted studies <strong>and</strong> outlines<br />
their pertinent features. The generalizability <strong>of</strong> these results to wider audiences <strong>and</strong> to lower risk<br />
populations has not been demonstrated, nor has the potential benefit for hospitalized patients<br />
been reported. Concerns with compliance could hinder their effectiveness on a population-wide<br />
level.<br />
Potential for Harm<br />
Discomfort from wearing the device, difficulty managing the garment while dealing with<br />
continence, <strong>and</strong> the potential for skin irritation <strong>and</strong> breakdown are causes for concern if fragile<br />
older people were to wear hip protectors. Because long-term compliance is low, it is unclear how<br />
many people would experience such problems if the devices were worn for longer periods during<br />
the day or for long-term use.<br />
296
Costs <strong>and</strong> Implementation<br />
An Australian study published in 2000 quoted a cost <strong>of</strong> A$10 per pair (approximately<br />
$5.25US). 4 The retail price quoted by one US manufacturer <strong>of</strong> a different hip protector is<br />
approximately $90 per pair. The lycra-containing undergarment used by some manufacturers to<br />
keep the hip pads in place requires special laundering <strong>and</strong> would require a tracking system<br />
similar to that used for other specialized garments or medical devices assigned to patients within<br />
a facility. Once provided, if devices can be put on <strong>and</strong> taken <strong>of</strong>f by individual users,<br />
implementation is straightforward. The cost-effectiveness <strong>of</strong> the devices has not been formally<br />
reported.<br />
Comment<br />
One <strong>of</strong> the main philosophical concerns raised by these studies is the change in emphasis<br />
from primary prevention <strong>of</strong> the underlying cause <strong>of</strong> hip fractures (ie, falls) to an emphasis on<br />
methods <strong>of</strong> protecting patients from the deleterious consequences <strong>of</strong> falls. However, a strategy<br />
for addressing the multiple risk factor model for falls is still warranted for primary falls<br />
prevention. With this caveat in mind, there is strong evidence to support the ability <strong>of</strong> hip<br />
protectors to prevent hip fractures. This evidence, in addition to their high face validity, may<br />
encourage their rapid adoption. Further evaluation <strong>of</strong> their costs, acceptability to patients, <strong>and</strong><br />
effectiveness in hospitalized patients (versus nursing home residents) is needed.<br />
297
Table 26.5.1. Hip protectors to prevent hip fracture*<br />
Study Participants <strong>and</strong><br />
Setting<br />
Parker,<br />
2000 9<br />
Chan,<br />
2000 4<br />
Ekman,<br />
1992 5<br />
Kannus,<br />
2000 6<br />
Lauritzen,<br />
1993 7<br />
Villar,<br />
1998 8<br />
1752 nursing home or<br />
rest home residents in<br />
5 countries<br />
71 nursing home<br />
residents in Australia,<br />
year not stated<br />
746 nursing home<br />
residents in Sweden,<br />
year not stated<br />
1801 community<br />
based elderly in<br />
Finl<strong>and</strong>, 1996-1997<br />
665 nursing home<br />
residents in Denmark,<br />
1991-1992<br />
141 rest home<br />
residents in the UK,<br />
year not stated<br />
Design,<br />
Outcomes<br />
Level 1A,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 3<br />
298<br />
Results<br />
* CI indicates confidence interval; OR, odds ratio; <strong>and</strong> RR, relative risk.<br />
References<br />
Peto OR 0.44 (95% CI: 0.26-0.75) <strong>of</strong> hip<br />
fracture in the intervention group in clusterr<strong>and</strong>omized<br />
studies;<br />
Peto OR 0.22 (95% CI: 0.09-0.57) in patientr<strong>and</strong>omized<br />
studies<br />
RR <strong>of</strong> hip fracture in the intervention group<br />
0.264 (95% CI: 0.073-0.959)<br />
RR <strong>of</strong> hip fracture in the intervention group<br />
0.33 (95% CI: 0.11-1.00)<br />
RR <strong>of</strong> hip fracture in the intervention group<br />
0.4 (95% CI: 0.2-0.8; p=0.008)<br />
RR <strong>of</strong> hip fracture in the intervention group<br />
0.44 (95% CI: 0.21-0.94)<br />
30% compliance over 3 months (hip fracture<br />
outcomes not assessed)<br />
1. Stevens JA, Olson S. Reducing falls <strong>and</strong> resulting hip fractures among older women.<br />
MMWR Morb Mortal Wkly Rep. 2000;49:1-12.<br />
2. Hannan EL, Magaziner J, Wang JJ, et al. Mortality <strong>and</strong> locomotion 6 months after<br />
hospitalization for hip fracture. JAMA. 2001;285:2736-2742.<br />
3. Magaziner J, Hawkes W, Hebel JR, et al. Recovery from hip fracture in eight areas <strong>of</strong><br />
function. J Gerontol. <strong>Medical</strong> Sciences 2000;55A:M498-M507.<br />
4. Chan DK, Hillier G, Coore M, et al. Effectiveness <strong>and</strong> acceptability <strong>of</strong> a newly designed<br />
hip protector: a pilot study. Arch Gerontol Geriatr. 2000;30:25-34.<br />
5. Ekman A, Mallmin H, Michaelsson K, Ljunghall S. External hip protectors to prevent<br />
osteoporotic hip fractures. Lancet. 1997;350:563-564.<br />
6. Kannus P, Parkkari J, Niemi S, et al. Prevention <strong>of</strong> hip fracture in elderly people with use<br />
<strong>of</strong> a hip protector. N Engl J Med. 2000;343:1506-1513.
7. Lauritzen JB, Petersen MM, Lund B. Effect <strong>of</strong> external hip protectors on hip fractures.<br />
Lancet 1993;341:11-13.<br />
8. Villar MTA, Hill P, Inskip H, Thompson P, Cooper C. Will elderly rest home residents<br />
wear hip protectors? Age Ageing. 1998;27:195-198.<br />
9. Parker MJ, Gillespie LD, Gillespie WJ. Hip protectors for preventing hip fractures in the<br />
elderly. In: The Cochrane Library, Issue 2, 2001. Oxford: Update S<strong>of</strong>tware.<br />
10. Harada A, Okulzumi H. Hip fracture prevention trial using hip protector in Japanese<br />
elderly. Osteoporos Int. 1998;8(Suppl 3):121.<br />
11. Heikinheimo R, Jantti PL, Aho HJ, Maki-Jokela PL. To fall but not to break – safety pants.<br />
Proceedings <strong>of</strong> the Third International Conference on Injury Prevention <strong>and</strong> Control,<br />
Melborne 1996;1996:576-578.<br />
299
300
Chapter 27. Prevention <strong>of</strong> Pressure Ulcers in Older <strong>Patient</strong>s<br />
Joseph V. Agostini, MD<br />
Dorothy I. Baker, PhD, RNCS<br />
Sidney T. Bogardus, Jr., MD<br />
Yale University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Pressure ulcers, localized areas <strong>of</strong> tissue damage or necrosis that develop due to pressure<br />
over a bony prominence, are common causes <strong>of</strong> morbidity in older hospitalized <strong>and</strong><br />
institutionalized persons. Other terms referring to the same phenomena are pressure sores,<br />
bedsores, <strong>and</strong> decubitus ulcers. Risk factors include immobility, friction, shear, incontinence,<br />
cognitive impairment <strong>and</strong> poor nutritional status. 1-3 Pressure ulcers are one indicator <strong>of</strong> quality <strong>of</strong><br />
care measured by nursing homes as part <strong>of</strong> the m<strong>and</strong>atory Minimum Data Set (MDS), which is<br />
required for participation in Medicare <strong>and</strong> Medicaid. Part <strong>of</strong> the MDS evaluation includes the<br />
Resident Assessment Instrument, which serves as a guide to assess pressure ulcers <strong>and</strong> many<br />
other pertinent clinical problems. 4<br />
Risk assessment is an integral part <strong>of</strong> prevention efforts. The Norton scale 5 <strong>and</strong> the<br />
Braden scale 6 are widely used tools to identify at-risk patients. The Norton scale assesses five<br />
domains: activity, incontinence, mental status, mobility, <strong>and</strong> physical condition. The Braden<br />
scale assesses six domains: activity, dietary intake, friction, mobility, sensory perception, <strong>and</strong><br />
skin moisture. Agreement between the scales is 0.73 using the kappa statistic. 7<br />
Different strategies have been used for primary prevention. Major clinical guidelines, 8 for<br />
pressure ulcer prevention are based primarily on published evidence, <strong>and</strong> in some areas, on<br />
pr<strong>of</strong>essional judgment <strong>and</strong> face validity <strong>of</strong> practices. Turning <strong>and</strong>/or repositioning patients is a<br />
practice with high face validity, but there are no well-designed controlled trials that examine its<br />
effect in the absence <strong>of</strong> other interventions. Other practices include regular skin inspection <strong>and</strong><br />
assessment, use <strong>of</strong> appropriate pressure-relief surfaces, improved mobility, adequate nutritional<br />
intake, <strong>and</strong> documentation <strong>of</strong> the skin examination. Additionally, the use <strong>of</strong> general educational<br />
interventions for hospital staff is supported by before-after study designs. 9, 10 Several reports<br />
suggest the value <strong>of</strong> using topical applications applied to intact skin in an attempt to prevent<br />
ulcers. 5, 11, 12 This chapter focuses on the use <strong>of</strong> pressure-relieving strategies that can be<br />
incorporated into hospital or nursing home practice <strong>and</strong> are based on evidence from controlled<br />
clinical trials.<br />
Practice Description<br />
The preventive practices that have received the most research attention are the use <strong>of</strong><br />
specific beds or mattresses. Many studies have compared a specific mattress with either another<br />
high-technology mattress or with a “st<strong>and</strong>ard” hospital mattress. A st<strong>and</strong>ard mattress, which is<br />
not uniformly defined in the literature, may be described as a hospital-issue, usually foam-based<br />
mattress found in a typical hospital room. The lack <strong>of</strong> consensus as to what constitutes a st<strong>and</strong>ard<br />
hospital mattress presents an interpretative challenge to investigators <strong>and</strong> administrators hoping<br />
to extrapolate results <strong>of</strong> a published trial to another locale.<br />
In 1991 Krasner reported that there were over 115 different pressure-relieving support<br />
surfaces on the market. 13 Sheepskin <strong>and</strong> other inexpensive pads (eg, “egg-crate” mattresses) are<br />
common pressure-reducing devices. Other static devices include pressure-relieving pads (eg,<br />
301
fabricated from elastic polymers) such as those used to cover operating room tables. Constant,<br />
low-pressure supports maintain uniform pressure throughout. Examples include higher-grade<br />
foam, <strong>and</strong> gel-, air-, or water-filled supports. In contrast, dynamic or alternating air devices have<br />
a built-in pump that continually redistributes air pressure. Low air-loss beds, as their name<br />
implies, permit small amounts <strong>of</strong> air to escape through a network <strong>of</strong> pores, whereas high air-loss<br />
(or air-fluidized) beds purposefully pump air through ceramic-type beads. Finally, kinetic<br />
turning beds, which allow continual rotation, are used more commonly in critically ill patients.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
In 1990 a large, prospective epidemiologic study reported the one-year incidence for<br />
pressure ulcer development in nursing homes to be 13.2%, 14 with prevalence reports ranging<br />
from 7% to 23% in a systematic review. 15 Risk-adjusted rates <strong>of</strong> new pressure ulcers have been<br />
reported to decrease by 25% from 1991 to 1996, based on a recent study using data from the<br />
Minimum Data Set. 16 In hospitalized patients, prevalence ranges from about 3% to 11%. 17<br />
Meehan reported that the prevalence <strong>of</strong> pressure ulcers was 11.1% in 177 hospitals surveyed in<br />
1993 <strong>and</strong> that 54% <strong>of</strong> the pressure ulcers occurred in patients 70 to 89 years old. 18 Eckman<br />
estimated that almost 1.7 million hospitalized patients had pressure ulcers. 19 Approimately 60%<br />
<strong>of</strong> pressure ulcers develop in these acute care settings. The National Health <strong>and</strong> Nutrition<br />
Examination Survey found that less than 20% <strong>of</strong> pressure ulcers arise in non-institutional<br />
environments. 20<br />
Pressure ulcers result in both increased length <strong>of</strong> hospital stay <strong>and</strong> hospital costs 21 <strong>and</strong><br />
increased nursing care time, as demonstrated by a study <strong>of</strong> long-term care patients. 22 Cellulitis,<br />
osteomyelitis, <strong>and</strong> sepsis are morbid complications <strong>of</strong> untreated pressure ulcers. Increased<br />
mortality has also been associated with pressure ulcers. 17<br />
Opportunities for Impact<br />
The passage <strong>of</strong> the Omnibus Budget Reconciliation Act <strong>of</strong> 1987 <strong>and</strong> subsequent<br />
implementation regulations provided a written m<strong>and</strong>ate for hospitalized <strong>and</strong> institutionalized<br />
patients to receive regular assessment, preventive measures, <strong>and</strong> treatment <strong>of</strong> pressure ulcers.<br />
There are no specific data on the current utilization <strong>of</strong> preventive measures by hospitals. The<br />
pressure ulcer protocols <strong>of</strong>ten vary from institution to institution.<br />
Study Designs<br />
We identified a recent systematic review <strong>of</strong> pressure-relieving devices 23 that evaluated 37<br />
r<strong>and</strong>omized controlled trials (RCTs), 31 <strong>of</strong> which were focused on pressure ulcer prevention<br />
(Table 27.1). Seven trials compared st<strong>and</strong>ard foam mattresses with various “low-technology”<br />
supports. These low-technology supports were defined as beds with constant low-pressure<br />
supports, including bead-, water-, or gel-filled supports; static air-filled supports; or foam or<br />
Silicore-filled supports. Seven <strong>of</strong> the trials compared constant low-pressure devices with<br />
alternating pressure devices. Six studies limited enrollment to orthopedic patients, 5 to patients<br />
in intensive care units, <strong>and</strong> 3 to patients undergoing operations (ie, the study evaluated different<br />
operating table surfaces). We identified no further RCTs <strong>of</strong> pressure ulcer prevention published<br />
after the systematic review identified above.<br />
Study Outcomes<br />
All studies reported pressure ulcer development in both intervention <strong>and</strong> control groups.<br />
Pressure ulcer grading systems were used in most trials. Typically, a 4-level grading system was<br />
302
employed, ranging from Grade 1 (discolored skin) to Grade 4 (full-thickness skin lesions with<br />
bone involvement).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Many specialized beds appear to be effective in reducing the development <strong>of</strong> pressure<br />
ulcers when compared with st<strong>and</strong>ard mattresses. For example, in 4 studies 24-27 cited by the<br />
systematic review 23 that compared st<strong>and</strong>ard hospital foam mattresses to enhanced foam<br />
alternatives, a summary relative risk <strong>of</strong> 0.29 (95% CI: 0.19-0.43) was calculated, favoring the<br />
intervention group. Between-group comparisons <strong>of</strong> the previously defined low-technology<br />
constant low-pressure devices did not yield clear conclusions. Similarly, in 7 RCTs the<br />
comparison <strong>of</strong> alternating pressure devices with a variety <strong>of</strong> constant low-pressure devices (a<br />
water mattress, foam pad, static air mattress, <strong>and</strong> foam overlays) showed no significant<br />
difference in pressure ulcer development (summary RR 0.84, 95% CI: 0.57-1.23). 23 However, a<br />
study <strong>of</strong> alternating pressure supports compared with st<strong>and</strong>ard foam mattresses did demonstrate<br />
lower pressure ulcer development in the intervention group (RR 0.32; 95% CI 0.14-0.74). 28<br />
Comparing pressure-reducing devices among themselves (versus against a st<strong>and</strong>ard mattress)<br />
yields no significant differences in the prevention <strong>of</strong> pressure ulcers. These trials have been<br />
summarized in a recent review. 29<br />
In addition, 2 published trials evaluating different operating table-like surfaces suggest<br />
reduction in pressure ulcer development with enhanced surfaces. 30, 31 In a well-designed RCT <strong>of</strong><br />
446 patients, Nixon <strong>and</strong> colleagues 31 showed that a dry gel-polymer pad placed on an operating<br />
table decreased the incidence <strong>of</strong> new pressure ulcers by almost half—11% for intervention<br />
patients vs. 20% for control patients placed on a st<strong>and</strong>ard operating table mattress (RR 0.46, 95%<br />
CI: 0.26-0.82).<br />
Several caveats temper the interpretation <strong>of</strong> studies <strong>of</strong> specialized pressure-relieving<br />
surfaces. In general the studies had poor methodologic design, as the systematic review points<br />
out. 23 The trials were mostly small, true baseline comparability was hard to confirm,<br />
st<strong>and</strong>ardization <strong>of</strong> protocols was <strong>of</strong>ten unclear, <strong>and</strong> assessments were frequently unblinded.<br />
<strong>Patient</strong> selection across trials was not consistent, <strong>and</strong> differences in pressure ulcer risk at<br />
enrollment were difficult to compare across studies.<br />
Potential for Harm<br />
None reported.<br />
Costs <strong>and</strong> Implementation<br />
The costs <strong>of</strong> treating a pressure ulcer are estimated to range from $4000 to $40,000 for<br />
newly developed ulcers. 32 Indeed, both hospital costs <strong>and</strong> length <strong>of</strong> stay are significantly higher<br />
for patients who develop pressure ulcers during hospitalization, as noted earlier. 21 In the nursing<br />
home in particular, failure to prevent this adverse outcome carries increasing liability—the<br />
median settlement for pressure ulcer-related disputes was $250,000 between the years 1977 <strong>and</strong><br />
1987. 33 The cost <strong>of</strong> specialized beds <strong>and</strong> mattresses to prevent pressure ulcer development can be<br />
high, ranging from $40 to $85 per day for low air-loss beds. 34 Specialized beds <strong>and</strong> intensive<br />
nursing interventions all carry clear resource implications. Inman <strong>and</strong> colleagues 35 have<br />
demonstrated the cost-effectiveness <strong>of</strong> an air suspension bed compared to a st<strong>and</strong>ard intensive<br />
care unit bed. Yet cost-effectiveness studies <strong>of</strong> the many different pressure-relieving devices<br />
have not been formally completed.<br />
303
In terms <strong>of</strong> the feasibility <strong>of</strong> implementing these specific devices <strong>and</strong> following<br />
guidelines for high-risk patients, both cost <strong>and</strong> time considerations must be examined. 36 Other<br />
considerations relate to the design <strong>and</strong> functionality <strong>of</strong> a particular bed or mattress—for example<br />
the ability <strong>of</strong> nursing staff to move <strong>and</strong> transfer patients placed on deeper or bulkier beds.<br />
Finally, difficulty in accurately assessing changes in the incidence <strong>and</strong> prevalence <strong>of</strong> pressure<br />
ulcers resulting from the institution <strong>of</strong> preventive measures is another barrier to documenting<br />
progress. 37<br />
Comment<br />
Overall there is adequate evidence that specially designed surfaces effectively prevent the<br />
development <strong>of</strong> pressure ulcers in high-risk patients, but the definition <strong>of</strong> high risk varies across<br />
studies. The variety <strong>of</strong> pressure-relieving devices makes it difficult to recommend one over<br />
another because there are few direct comparisons among the many different types <strong>of</strong> surfaces. Of<br />
note, the treatment <strong>of</strong> established pressure ulcers is a separate topic, <strong>and</strong> the type <strong>of</strong> pressurerelieving<br />
surface that is effective in treatment may not prove best for prevention. Appropriate<br />
patient selection criteria need further development <strong>and</strong> refinement because the cost <strong>of</strong> many<br />
prevention interventions is high. The necessity for larger RCTs to assess both clinical <strong>and</strong> costeffectiveness<br />
<strong>of</strong> these specially designed mattresses is clear. Better descriptions <strong>of</strong> what<br />
constitutes a “st<strong>and</strong>ard” bed or mattress, <strong>and</strong> improved reporting <strong>of</strong> baseline comparability<br />
between experimental <strong>and</strong> control groups are also necessary to adequately interpret existing<br />
studies. To better track progress in prevention, st<strong>and</strong>ardized strategies should be developed so<br />
that accurate prevalence estimates can be documented.<br />
Table 27.1. Studies <strong>of</strong> pressure ulcer prevention*<br />
Participants <strong>and</strong> Setting Study Design,<br />
Outcomes<br />
Systematic review <strong>of</strong> 31 RCTs from<br />
the US, UK <strong>and</strong> elsewhere assessing<br />
pressure relieving interventions for<br />
prevention <strong>of</strong> pressure ulcers 23<br />
Level 1A,<br />
Level 1<br />
304<br />
Relative Risk <strong>of</strong> Pressure Ulcer<br />
Enhanced alternative foam mattresses<br />
vs. st<strong>and</strong>ard hospital mattresses:<br />
RR 0.29 (95% CI: 0.19-0.43)<br />
Alternating pressure vs. constant low<br />
pressure devices: RR 0.84<br />
(95% CI: 0.57-1.23)<br />
* CI indicates confidence interval; RCT, r<strong>and</strong>omized controlled trial; <strong>and</strong> RR, relative risk.<br />
References<br />
1. Berlowitz DR, Wilking SV. Risk factors for pressure sores. A comparison <strong>of</strong> crosssectional<br />
<strong>and</strong> cohort-derived data. J Am Geriatr Soc. 1989;37:1043-1050.<br />
2. Br<strong>and</strong>eis GH, Ooi WL, Hossain M, Morris JN, Lipsitz LA. A longitudinal study <strong>of</strong> risk<br />
factors associated with the formation <strong>of</strong> pressure ulcers in nursing homes. J Am Geriatr<br />
Soc. 1994;42:388-393.<br />
3. Allman RM, Goode PS, Patrick MM, Burst N, Bartolucci AA. Pressure ulcer risk factors<br />
among hospitalized patients with activity limitation. JAMA. 1995;273:865-870.<br />
4. Long term care facility resident assessment (RAI) user's manual. Version 2.0. Baltimore,<br />
MD: Health Care Financing Administration; 1995.
5. Norton D, McLaren R, Exton-Smith AN. An investigation <strong>of</strong> geriatric nursing problems in<br />
hospital. London: Churchill Livingstone; 1962.<br />
6. Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden Scale for Predicting Pressure<br />
Sore Risk. Nurs Res. 1987;36:205-210.<br />
7. Xakellis GC, Frantz RA, Arteaga M, Nguyen M, Lewis A. A comparison <strong>of</strong> patient risk for<br />
pressure ulcer development with nursing use <strong>of</strong> preventive interventions. J Am Geriatr Soc.<br />
1992;40:1250-1254.<br />
8. Agency for Health Care Policy <strong>and</strong> Research. Pressure Ulcer in Adults: Prediction <strong>and</strong><br />
Prevention. Clinical Practice Guideline, Number 3. Rockville, MD: <strong>Public</strong> Health Service,<br />
U.S. Department <strong>of</strong> Health <strong>and</strong> Human Services; 1992.<br />
9. Blom MF. Dramatic decrease in decubitus ulcers. Geriatr Nurs. 1985;6:84-87.<br />
10. Moody BL, Fanale JE, Thompson M, Vaillancourt D, Symonds G, Bonasoro C. Impact <strong>of</strong><br />
staff education on pressure sore development in elderly hospitalized patients. Arch Intern<br />
Med. 1988;148:2241-2243.<br />
11. van der Cammen TJ, O'Callaghan U, Whitefield M. Prevention <strong>of</strong> pressure sores. A<br />
comparison <strong>of</strong> new <strong>and</strong> old pressure sore treatments. Br J Clin Pract 1987;41:1009-1011.<br />
12. Declair V. The usefulness <strong>of</strong> topical application <strong>of</strong> essential fatty acids (EFA) to prevent<br />
pressure ulcers. Ostomy Wound Management. 1997;43:48-52,54.<br />
13. Krasner D. <strong>Patient</strong> support surfaces. Ostomy Wound Management. 1991;33:47-56.<br />
14. Br<strong>and</strong>eis GH, Morris JN, Nash DJ, Lipsitz LA. The epidemiology <strong>and</strong> natural history <strong>of</strong><br />
pressure ulcers in elderly nursing home residents. JAMA. 1990;264:2905-2909.<br />
15. Smith DM. Pressure ulcers in the nursing home. Ann Intern Med. 1995;123:433-442.<br />
16. Berlowitz DR, Bezerra HQ, Br<strong>and</strong>eis GH, Kader B, Anderson JJ. Are we improving the<br />
quality <strong>of</strong> nursing home care: the case <strong>of</strong> pressure ulcers. J Am Geriatr Soc. 2000;48:59-62.<br />
17. Allman RM, Laprade CA, Noel LB, Walker JM, Moorer CA, Dear MR, et al. Pressure<br />
sores among hospitalized patients. Ann Intern Med. 1986;105:337-342.<br />
18. Meehan M. National pressure ulcer prevalence survey. Adv Wound Care. 1994;7:27-<br />
30,34,36-8.<br />
19. Eckman KL. The prevalence <strong>of</strong> dermal ulcers among persons in the U.S. who have died.<br />
Decubitus. 1989;2:36-40.<br />
20. Guralnik JM, Harris TB, White LR, Cornoni-Huntley JC. Occurrence <strong>and</strong> predictors <strong>of</strong><br />
pressure sores in the National Health <strong>and</strong> Nutrition Examination survey follow-up. J Am<br />
Geriatr Soc. 1988;36:807-812.<br />
21. Allman RM, Goode PS, Burst N, Bartolucci AA, Thomas DR. Pressure ulcers, hospital<br />
complications, <strong>and</strong> disease severity: impact on hospital costs <strong>and</strong> length <strong>of</strong> stay. Adv<br />
Wound Care. 1999;12:22-30.<br />
22. D'Hoore W, Guisset AL, Tilquin C. Increased nursing time requirements due to pressure<br />
sores in long-term-care residents in Quebec. Clin Perform Qual Health Care. 1997;5:189-<br />
194.<br />
23. Cullum N, Deeks J, Sheldon TA, Song F, Fletcher AW. Beds, mattresses <strong>and</strong> cushions for<br />
pressure sore prevention <strong>and</strong> treatment. In: The Cochrane Library, Issue 4, 2000. Oxford:<br />
Update S<strong>of</strong>tware.<br />
24. H<strong>of</strong>man A, Geelkerken RH, Wille J, Hamming JJ, Hermans J, Breslau PJ. Pressure sores<br />
<strong>and</strong> pressure-decreasing mattresses: controlled clinical trial. Lancet. 1994;343:568-571.<br />
25. Santy JE, Butler MK, Whyman JD. A comparison study <strong>of</strong> 6 types <strong>of</strong> hospital mattress to<br />
determine which most effectively reduces the incidence <strong>of</strong> pressure sores in elderly patients<br />
305
with hip fractures in a District General Hospital. Report to Northern & Yorkshire Regional<br />
Health Authority 1994.<br />
26. Gray DG, Campbell M. A r<strong>and</strong>omized clinical trial <strong>of</strong> two types <strong>of</strong> foam mattresses. J<br />
Tissue Viability. 1994;4:128-132.<br />
27. Collier ME. Pressure-reducing mattresses. J Wound Care. 1996;5:207-211.<br />
28. Andersen KE, Jensen O, Kvorning SA, Bach E. Decubitus prophylaxis: a prospective trial<br />
on the efficiency <strong>of</strong> alternating-pressure air-mattresses <strong>and</strong> water-mattresses. Acta Derm<br />
Venereol (Stockholm). 1982;63:227-230.<br />
29. Thomas DR. Issues <strong>and</strong> dilemmas in the prevention <strong>and</strong> treatment <strong>of</strong> pressure ulcers: a<br />
review. J Gerontol. 2001;56A:M328-M40.<br />
30. Aronovitch SA. A comparative, r<strong>and</strong>omized, controlled study to determine safety <strong>and</strong><br />
efficacy <strong>of</strong> preventive pressure ulcer systems: preliminary analysis. Adv Wound Care.<br />
1998;11:15-16.<br />
31. Nixon J, McElvenny D, Mason S, Brown J, Bond S. A sequential r<strong>and</strong>omised controlled<br />
trial comparing a dry visco-elastic polymer pad <strong>and</strong> st<strong>and</strong>ard operating table mattress in the<br />
prevention <strong>of</strong> post-operative pressure sores. Int J Nurs Stud. 1998;35:193-203.<br />
32. Hibbs P. The economics <strong>of</strong> pressure ulcer prevention. Decubitus. 1988;1:32-38.<br />
33. Bennett RG, O'Sullivan J, DeVito EM, Remsburg R. The increasing medical malpractice<br />
risk related to pressure ulcers in the United States. J Am Geriatr Soc. 2000;48:73-81.<br />
34. Ferrell BA, Keeler E, Siu AL, Ahn S, Osterweil D. Cost-effectiveness <strong>of</strong> low-air-loss beds<br />
for treatment <strong>of</strong> pressure ulcers. J Gerontol. 1995;50A:M141-M46.<br />
35. Inman KJ, Sibbald WJ, Rutledge FS, Clark BJ. Clinical utility <strong>and</strong> cost-effectiveness <strong>of</strong> an<br />
air suspension bed in the prevention <strong>of</strong> pressure ulcers. JAMA. 1993;269:1139-1143.<br />
36. Hu TW, Stotts NA, Fogarty TE, Bergstrom N. Cost analysis for guideline implementation<br />
in prevention <strong>and</strong> early treatment <strong>of</strong> pressure ulcers. Decubitus. 1993;6:42-46.<br />
37. Smith DM, Winsemius DK, Besdine RW. Pressure sores in the elderly: can this outcome<br />
be improved? J Gen Intern Med. 1991;6:81-93.<br />
306
Chapter 28. Prevention <strong>of</strong> Delirium in Older Hospitalized <strong>Patient</strong>s<br />
Joseph V. Agostini, MD<br />
Dorothy I. Baker, PhD, RNCS<br />
Sharon K. Inouye, MD, MPH<br />
Sidney T. Bogardus, Jr., MD<br />
Yale University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Delirium, or acute confusional state, is a common complication among hospitalized older<br />
patients. Delirium is characterized by a sudden onset <strong>and</strong> fluctuating course, inattention, altered<br />
level <strong>of</strong> consciousness, disorganized thought <strong>and</strong> speech, disorientation, <strong>and</strong> <strong>of</strong>ten behavioral<br />
disturbance. As with other common geriatric syndromes, the etiology <strong>of</strong> delirium is<br />
multifactorial. Previous research has identified a broad range <strong>of</strong> predisposing <strong>and</strong> precipitating<br />
factors. 1-4 These include older age, cognitive or sensory impairments, dehydration, specific<br />
medication usage (eg, psychoactive drugs), concurrent medical illness, <strong>and</strong> sleep deprivation.<br />
The multifactorial nature <strong>of</strong> delirium suggests that intervention strategies targeting multiple<br />
known risk factors might be effective in preventing its occurrence in hospitalized older patients.<br />
In this chapter, we review multicomponent prevention programs that can be applied to a general<br />
hospitalized patient population, not restricted to one admitting diagnosis (in keeping with the<br />
crosscutting patient safety focus <strong>of</strong> the Report; Chapter 1). For example, a study comparing the<br />
effect <strong>of</strong> postoperative analgesia using intravenous versus epidural infusions after bilateral knee<br />
replacement surgery was not included. 5<br />
Practice Description<br />
A number <strong>of</strong> individual interventions have been used in efforts to prevent delirium. Some<br />
could be considered part <strong>of</strong> general nursing practice, whereas others involve medical<br />
assessments by physicians or consultants. General strategies to prevent delirium include use <strong>of</strong><br />
patient reorientation techniques (such as verbal reassurance, re-introduction <strong>of</strong> team members,<br />
review <strong>of</strong> the daily hospital routine <strong>and</strong> patient schedule), environmental modifications (visible<br />
clocks <strong>and</strong> calendars), <strong>and</strong> scheduled patient mobility. The number <strong>and</strong> complexity <strong>of</strong> these<br />
interventions can vary, with individual nursing discretion usually determining how <strong>and</strong> when<br />
these interventions are implemented. <strong>Patient</strong> education, 6 nursing staff education, 7 <strong>and</strong> family<br />
involvement 8 are also useful. Approaches for primary prevention that incorporate physician<br />
consultants or geriatric consultative teams 9-11 are reviewed elsewhere in this Report (see<br />
Chapters 29 <strong>and</strong> 30).<br />
Formal prevention programs target defined risk factors by implementing multiple<br />
practices according to st<strong>and</strong>ardized protocols. For example, a recently reported, multicomponent<br />
strategy focused on 6 risk factors <strong>and</strong> successfully developed intervention protocols to address<br />
each <strong>of</strong> them. 12 <strong>Patient</strong>s with cognitive impairment received daily orientation interventions <strong>and</strong><br />
3-times daily cognitive stimulation activities. To target sleep impairment, patients received nonpharmacologic<br />
sleeping aids (eg, back massage <strong>and</strong> relaxation tapes), while hospital staff<br />
engaged in noise-reduction strategies such as setting beepers to vibrate <strong>and</strong> using silent pill<br />
crushers. Immobility was addressed with a 3-times daily exercise protocol adapted for use with<br />
bed-bound <strong>and</strong> ambulatory patients. Sensory impairments were addressed by providing devices<br />
such as auditory amplifiers, visual aids, <strong>and</strong> larger size push-button phones. <strong>Patient</strong>s with<br />
307
evidence <strong>of</strong> dehydration received st<strong>and</strong>ardized repletion interventions. A geriatric nurse<br />
specialist <strong>and</strong> staff assisted by trained volunteers carried out all the interventions.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The target safety problem is the primary prevention <strong>of</strong> delirium, rather than the<br />
treatment 13 <strong>of</strong> existing delirium. In the United States, delirium affects an estimated 2.3 million<br />
hospitalized elders annually, accounting for 17.5 million inpatient days, <strong>and</strong> leading to more than<br />
$4 billion in Medicare costs (1994 dollars). 12 Studies have found that delirium in hospitalized<br />
patients contributes to longer lengths <strong>of</strong> stay, 14 increased mortality, 15-17 <strong>and</strong> increased rates <strong>of</strong><br />
institutional placement. 18, 19 New cases <strong>of</strong> delirium occur in approximately 15% to 60% <strong>of</strong><br />
hospitalized older patients, depending on the number <strong>of</strong> risk factors present at<br />
admission. 4,15,18,20,21 Moreover, because many cases <strong>of</strong> delirium go unrecognized during<br />
hospitalization <strong>and</strong> because symptoms may persist for months after discharge, 22 these may be<br />
conservative estimates. <strong>Safety</strong> practices to reduce delirium may thus have substantial impact on<br />
the health <strong>and</strong> well-being <strong>of</strong> older patients in hospitals. These practices may also impact nursing<br />
home residents <strong>and</strong> other institutionalized patients, but our practice review did not identify any<br />
studies carried out among these patient populations.<br />
Opportunities for Impact<br />
It is difficult to estimate the extent <strong>of</strong> existing practices aimed at decreasing delirium. A<br />
comprehensive model, the Hospital Elder Life Program, 23 which incorporates the delirium<br />
interventions reviewed in one study in this chapter, 12 is presently in the initial dissemination<br />
phase at 6 replication sites, with 16 hospitals on a waiting list. Present evidence suggests that few<br />
facilities currently have intervention programs designed for the primary prevention <strong>of</strong> delirium.<br />
The opportunity for impact in nursing homes <strong>and</strong> other long-term care facilities is great, but thus<br />
far studies have not targeted these settings.<br />
Study Designs<br />
Cole 24 conducted a structured search <strong>of</strong> the medical literature <strong>and</strong> identified 10<br />
intervention trials to prevent delirium in hospitalized patients. Of these, we excluded one study<br />
<strong>of</strong> much younger patients (mean age, 49 years) 25 <strong>and</strong> one study that incorporated interventions<br />
not applicable to most hospitalized elders (eg, early surgery, prevention <strong>and</strong> treatment <strong>of</strong> perioperative<br />
blood pressure falls). 26 Three used psychiatric consultations 27-29 which did not fit our<br />
criteria for risk factor intervention (see Chapter 29 for similar studies). Table 28.1 lists the<br />
remaining 5 studies 6, 8, 30-32 <strong>and</strong> a later study, 12 which is the largest controlled trial to date.<br />
Study Outcomes<br />
All <strong>of</strong> the studies in Table 28.1 reported delirium or confusion symptoms as an outcome<br />
measure. Each study, however, used a different instrument to identify delirium: DSM-III, 33 the<br />
Confusion Assessment Method, 34 the Short Portable Mental Status Questionnaire, 35 or a scoring<br />
system based on delirium symptoms.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The earliest studies, by Owens 6 <strong>and</strong> Chatham, 8 focused on the effects <strong>of</strong> patient <strong>and</strong><br />
family education, respectively. Delirium symptoms modestly improved but achieved statistical<br />
significance in only 5 <strong>of</strong> the 11 symptom categories reported in the latter study. Both studies<br />
were limited by small numbers <strong>of</strong> patients, non-st<strong>and</strong>ardized interventions, <strong>and</strong> minimal data on<br />
308
aseline co-morbidities <strong>of</strong> the enrolled patients. The study by Williams <strong>and</strong> colleagues, 32 which<br />
targeted a population at high risk for delirium (older patients with hip fracture), also<br />
demonstrated a statistically significant reduction in delirium symptoms by targeting<br />
environmental nursing interventions <strong>and</strong> patient education. Two subsequent studies did not show<br />
a reduction in delirium. The low incidence <strong>of</strong> delirium (only 3 cases in 30 intervention patients)<br />
in the study by Nagley et al 30 created inadequate power to detect a significant effect with only 60<br />
total patients. Although a high percentage <strong>of</strong> patients experienced delirium in the study by<br />
Wanich et al, 31 79% <strong>of</strong> cases were diagnosed at the time <strong>of</strong> admission (prevalent rather than<br />
incident cases) <strong>and</strong> therefore could not have been prevented by the intervention. Both <strong>of</strong> these<br />
studies may also have suffered from contamination bias. The greatest benefit in delirium<br />
prevention, a 40% risk reduction, occurred in the study by Inouye et al, 12 a carefully designed<br />
<strong>and</strong> implemented hospital program targeting 6 well-recognized risk factors for delirium, in<br />
which adherence to each intervention protocol was tracked. The intervention reduced the number<br />
<strong>and</strong> severity <strong>of</strong> patients’ risk factors <strong>and</strong> was successful in preventing patients’ first delirium<br />
episode.<br />
Potential for Harm<br />
None noted.<br />
Costs <strong>of</strong> Implementation<br />
The only recent estimate <strong>of</strong> cost per case <strong>of</strong> delirium prevented was $6341 in a delirium<br />
prevention trial, 12 which is less than the cost associated with prevention <strong>of</strong> other hospital<br />
complications such as falls. A further analysis <strong>of</strong> the same patients reveals that the<br />
multicomponent strategy is cost-effective for those at intermediate risk <strong>of</strong> delirium, but not for<br />
those at highest risk. 36<br />
Comment<br />
The literature for delirium prevention studies is small, <strong>and</strong> the methodologic quality <strong>of</strong><br />
many studies is poor. However, one high quality study 12 has demonstrated that multicomponent<br />
interventions can prevent incident delirium in hospitalized patients. The interventions have high<br />
face validity <strong>and</strong> are both feasible <strong>and</strong> transportable across institutions <strong>and</strong> hospital units,<br />
suggesting that implementation in different practice settings would be practical. Implementing a<br />
multicomponent intervention on a hospital-wide basis throughout the United States would<br />
require significant commitment from hospital staff. Programs such as the Hospital Elder Life<br />
Program 23 can be readily integrated into hospital practice <strong>and</strong> have been successful in preventing<br />
both cognitive <strong>and</strong> functional decline using targeted, practical interventions. Others <strong>of</strong> these<br />
practices could be incorporated by either support staff or trained volunteers, which may save<br />
resources <strong>and</strong> underscore the fact that many common sense interventions do not require a larger<br />
pr<strong>of</strong>essional staff. Future studies should focus on refining the most effective multifactorial<br />
programs, determining the optimal combination <strong>of</strong> interventions, defining appropriate target<br />
populations based on delirium risk, demonstrating effectiveness across multiple clinical sites,<br />
<strong>and</strong> disseminating the most cost-effective practices.<br />
309
Table 28.1. Six studies <strong>of</strong> delirium prevention*<br />
Study Study Setting Interventions Study Design<br />
Outcomes<br />
Chatham,<br />
1978 8<br />
Inouye,<br />
1999 12<br />
Nagley,<br />
1986 30<br />
Owens,<br />
1982 6<br />
Wanich,<br />
1992 31<br />
Williams,<br />
1985 32<br />
20 surgical<br />
patients in a<br />
university<br />
affiliated<br />
hospital, 1977<br />
852 patients in a<br />
university<br />
hospital, 1995-<br />
1998<br />
60 patients at a<br />
university<br />
affiliated<br />
hospital<br />
64 surgical<br />
patients in a<br />
university<br />
hospital<br />
235 patients in a<br />
university<br />
hospital, 1986-<br />
1987<br />
227 orthopedic<br />
patients in 4<br />
hospitals<br />
• Family education<br />
• <strong>Patient</strong> education<br />
Targeted 6 risk factors:<br />
• Cognitive impairment<br />
• Immobility<br />
• Visual impairment<br />
• Hearing impairment<br />
• Dehydration<br />
• Sleep deprivation<br />
16 interventions, including:<br />
• Orientation strategies<br />
• Providing sensory aides<br />
• Ambulation<br />
• Hydration measures<br />
• Nursing interaction<br />
310<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
• <strong>Patient</strong> education Level 2,<br />
Level 1<br />
• Nursing education<br />
• Caregiver education<br />
• Orientation strategies<br />
• Mobilization<br />
• Environmental<br />
modifications<br />
• Medication evaluation<br />
• <strong>Patient</strong> education<br />
• Orientation strategies<br />
• Providing sensory aides<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
* CI indicates confidence interval; OR, odds ratio.<br />
† Delirium rate is the percentage <strong>of</strong> patients with one or more episodes <strong>of</strong> delirium.<br />
Results†<br />
Delirium symptoms rate:<br />
intervention resulted in<br />
improvement in 5 <strong>of</strong> 11<br />
areas—orientation,<br />
appropriateness, confusion,<br />
delusions, <strong>and</strong> sleep (p0.05)<br />
Delirium symptoms rate:<br />
intervention 59%, control<br />
78% (p>0.05)<br />
Delirium rate: intervention<br />
19%, control 22% (p=0.61)<br />
Delirium symptoms rate:<br />
intervention 43.9%, control<br />
51.5% (p
References<br />
1. Elie M, Cole MG, Primeau FJ, Bellavance F. Delirium risk factors in elderly hospitalized<br />
patients. J Gen Intern Med. 1998;13:204-212.<br />
2. Inouye SK, Viscoli CM, Horwitz RI, Hurst LD, Tinetti ME. A predictive model for<br />
delirium in hospitalized elderly medical patients based on admission characteristics. Ann<br />
Intern Med. 1993;119:474-481.<br />
3. Schor JD, Levk<strong>of</strong>f SE, Lipsitz LA, Reilly CH, Cleary PD, Rowe JW, et al. Risk factors for<br />
delirium in hospitalized elderly. JAMA. 1992;267:827-831.<br />
4. Inouye SK, Charpentier PA. Precipitating factors for delirium in hospitalized elderly<br />
persons. Predictive model <strong>and</strong> interrelationship with baseline vulnerability. JAMA.<br />
1996;275:852-857.<br />
5. Williams-Russo P, Urquhart BL, Sharrock NE, Charlson ME. Post-operative delirium:<br />
predictors <strong>and</strong> prognosis in elderly orthopedic patients. J Am Geriatr Soc. 1992;40:759-<br />
767.<br />
6. Owens JF, Hutelmyer CM. The effect <strong>of</strong> preoperative intervention on delirium in cardiac<br />
surgical patients. Nurs Res. 1982;31:60-62.<br />
7. Rapp CG, Onega LL, Tripp-Reimer T, Mobily P, Wakefield B, Kundrat M, et al. Unitbased<br />
acute confusion resource nurse: an educational program to train staff nurses.<br />
Gerontologist. 1998;38:628-632.<br />
8. Chatham MA. The effect <strong>of</strong> family involvement on patients' manifestations <strong>of</strong><br />
postcardiotomy psychosis. Heart Lung. 1978;7:995-999.<br />
9. Hogan DB, Fox RA, Gadley BW, Mann OE. Effect <strong>of</strong> a geriatric consultation service on<br />
management <strong>of</strong> patients in an acute care hospital. CMAJ. 1987;136:713-717.<br />
10. Cole MG, Fenton FR, Englesmann F, Mansouri I. Effectiveness <strong>of</strong> geriatric psychiatry<br />
consultation in an acute care hospital: a r<strong>and</strong>omized clinical trial. J Am Geriatr Soc.<br />
1991;39:1183-1188.<br />
11. Marcantonio ER, Flacker JM, Wright RJ, Resnick NM. Reducing delirium after hip<br />
fracture: a r<strong>and</strong>omized trial. J Am Geriatr Soc. 2001;49:In press.<br />
12. Inouye SK, Bogardus ST, Jr., Charpentier PA, Leo-Summers L, Acampora D, Holford TR,<br />
et al. A multicomponent intervention to prevent delirium in hospitalized older patients. N<br />
Engl J Med. 1999;340:669-676.<br />
13. Meagher DJ. Delirium: optimising management. BMJ. 2001; 322:144-149.<br />
14. Pompei P, Foreman M, Rudberg MA, Inouye SK, Braund V, Cassel CK. Delirium in<br />
hospitalized older persons: outcomes <strong>and</strong> predictors. J Am Geriatr Soc. 1994;42:809-815.<br />
15. Cameron DJ, Thomas RI, Mulvihill M, Bronheim H. Delirium: a test <strong>of</strong> the Diagnostic <strong>and</strong><br />
Statistical Manual III criteria on medical inpatients. J Am Geriatr Soc. 1987;35:1007-1010.<br />
16. Weddington WW. The mortality <strong>of</strong> delirium: an underappreciated problem?<br />
Psychosomatics. 1982;23:1232-1235.<br />
17. Cole MG, Primeau FJ. Prognosis <strong>of</strong> delirium in elderly hospital patients. CMAJ.<br />
1993;149:41-46.<br />
18. Francis J, Martin D, Kapoor WN. A prospective study <strong>of</strong> delirium in hospitalized elderly.<br />
JAMA. 1990;263:1097-1101.<br />
19. Inouye SK, Rushing JT, Foreman MD, Palmer RM, Pompei P. Does delirium contribute to<br />
poor hospital outcomes? A three-site epidemiologic study. J Gen Intern Med. 1998;13:234-<br />
242.<br />
311
20. Foreman MD. Confusion in the hospitalized elderly: incidence, onset, <strong>and</strong> associated<br />
factors. Res Nurs Health. 1989;12:21-29.<br />
21. Rockwood K. Acute confusion in elderly medical patients. J Am Geriatr Soc. 1989;37:150-<br />
154.<br />
22. Levk<strong>of</strong>f SE, Evans DA, Liptzin B, Cleary PD, Lipsitz LA, Wetle TT, et al. Delirium. The<br />
occurrence <strong>and</strong> persistence <strong>of</strong> symptoms among elderly hospitalized patients. Arch Intern<br />
Med. 1992;152:334-340.<br />
23. Inouye SK, Bogardus ST, Jr., Baker DI, Leo-Summers L, Cooney LM, Jr. The Hospital<br />
Elder Life Program: a model <strong>of</strong> care to prevent cognitive <strong>and</strong> functional decline in older<br />
hospitalized patients. J Am Geriatr Soc. 2000;48:1697-1706.<br />
24. Cole MG, Primeau F, McCusker J. Effectiveness <strong>of</strong> interventions to prevent delirium in<br />
hospitalized patients: a systematic review. CMAJ. 1996;155:1263-1268.<br />
25. Budd S, Brown W. Effect <strong>of</strong> a reorientation technique on postcardiotomy delirium. Nurs<br />
Res. 1974;23:341-348.<br />
26. Gustafson Y, Brannstrom B, Berggren D, Ragnarsson JI, Sigaard J, Bucht G, et al. A<br />
geriatric-anesthesiologic program to reduce acute confusional states in elderly patients<br />
treated for femoral neck fractures. J Am Geriatr Soc. 1991;39:6556-62.<br />
27. Layne OL, Jr., Yud<strong>of</strong>sky SC. Postoperative psychosis in cardiotomy patients. The role <strong>of</strong><br />
organic <strong>and</strong> psychiatric factors. N Engl J Med. 1971;284:518-520.<br />
28. Schindler BA, Shook J, Schwartz GM. Beneficial effects <strong>of</strong> psychiatric intervention on<br />
recovery after coronary artery bypass graft surgery. Gen Hosp Psychistry. 1989;11:358-<br />
364.<br />
29. Lazarus HR, Hagens JH. Prevention <strong>of</strong> psychosis following open-heart surgery. Am J<br />
Psychiatry. 1968;124:1190-1195.<br />
30. Nagley SJ. Predicting <strong>and</strong> preventing confusion in your patients. J Gerontol Nurs.<br />
1986;12:27-31.<br />
31. Wanich CK, Sullivan-Marx EM, Gottlieb GL, Johnson JC. Functional status outcomes <strong>of</strong> a<br />
nursing intervention in hospitalized elderly. Image J Nurs Sch. 1992;24:201-207.<br />
32. Williams MA, Campbell EB, Raynor WJ, Mlynarczyk SM, Ward SE. Reducing acute<br />
confusional states in elderly patients with hip fractures. Res Nurs Health. 1985;8:329-337.<br />
33. American Psychiatric Association. Diagnostic <strong>and</strong> statistical manual <strong>of</strong> mental disorders<br />
(3rd edition). ed. Washington, DC: American Psychiatric Press; 1980.<br />
34. Inouye SK, van Dyck CH, Alessi CA, Balkin S, Siegal AP, Horwitz RI. Clarifying<br />
confusion: the confusion assessment method. A new method for detection <strong>of</strong> delirium. Ann<br />
Intern Med. 1990;113:941-948.<br />
35. Pfeiffer E. A short portable mental status questionnaire for assessment <strong>of</strong> organic brain<br />
deficit in elderly patients. J Am Geriatr Soc. 1979;23:433-438.<br />
36. Rizzo JA, Bogardus ST, Jr., Leo-Summers L, Williams CS, Acampora D, Inouye SK. A<br />
multicomponent targeted intervention to prevent delirium in hospitalized older patients:<br />
what is the economic value? <strong>Medical</strong> Care. 2001;In press.<br />
312
Chapter 29. Multidisciplinary Geriatric Consultation Services<br />
Joseph V. Agostini, MD<br />
Dorothy I. Baker, PhD, RNCS<br />
Sharon K. Inouye, MD, MPH<br />
Sidney T. Bogardus, Jr., MD<br />
Yale University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Multidisciplinary geriatric consultation teams provide comprehensive assessment <strong>of</strong><br />
physical, emotional, <strong>and</strong> functional status in older persons <strong>and</strong> make recommendations regarding<br />
prevention <strong>and</strong> management <strong>of</strong> common geriatric syndromes, functional impairments, <strong>and</strong> other<br />
problems. Teams vary in composition but usually include a geriatrician, nurse, social worker <strong>and</strong><br />
other health pr<strong>of</strong>essionals such as rehabilitative therapists, psychologists, <strong>and</strong> dieticians. Their<br />
expertise may improve the safety <strong>of</strong> hospitalized elders (or nursing home residents) by reducing<br />
hospital-acquired complications such as falls, delirium, functional decline, <strong>and</strong> “preventable”<br />
deaths. Consultation teams <strong>and</strong> inpatient geriatric evaluation <strong>and</strong> management (GEM) units share<br />
this multidisciplinary approach but differ in who controls the implementation <strong>of</strong><br />
recommendations. A patient’s primary physician decides whether to implement a consultation<br />
team’s recommendations, whereas geriatric teams have direct control over implementation in<br />
GEM units. As this difference may impact effectiveness, GEM units are reviewed separately<br />
(Chapter 30). Multidisciplinary consultation services are available to elders living in the<br />
community, however this chapter reviews the evidence for geriatric consultation improving the<br />
safety <strong>of</strong> hospitalized patients or nursing home residents.<br />
Practice Description<br />
The structure <strong>of</strong> consultation teams, the types <strong>of</strong> evaluations they routinely perform, <strong>and</strong><br />
the recommendations they make vary from institution to institution. A review <strong>of</strong> published<br />
studies reveals some common features. A team has at least one physician, a nurse practitioner or<br />
similarly pr<strong>of</strong>essionally-trained provider, rehabilitative experts, <strong>and</strong> usually a social worker.<br />
Assessments typically include measures <strong>of</strong> mobility <strong>and</strong> functional status, mental status<br />
examinations, psychological screening, evaluation <strong>of</strong> common geriatric problems (eg, risk for<br />
falls, incontinence, <strong>and</strong> polypharmacy), <strong>and</strong> plans for rehabilitation <strong>and</strong>/or appropriate placement<br />
at the time <strong>of</strong> discharge. The team notes its recommendations in the hospital chart,<br />
communicates them to the physician directing care for the patient, <strong>and</strong> provides follow-up until<br />
the patient is discharged.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
<strong>Patient</strong>s aged 65 years <strong>and</strong> older account for almost half <strong>of</strong> inpatient hospitalization<br />
days. 1 In 1996, they comprised 13% <strong>of</strong> the population yet accounted for 38% <strong>of</strong> the<br />
approximately 31 million discharges from non-government, acute care hospitals. 2 The actual<br />
target population is smaller because, although all hospitalized elders are at risk for<br />
complications, some patients are unlikely to benefit from multidisciplinary geriatric consultation.<br />
Strategies to target geriatric services to those patients most likely to benefit have been reviewed<br />
in the literature. 3 The characteristics associated with potential benefit include advanced age (eg,<br />
over 75 years old), specific geriatric conditions (eg, falls or confusion), functional impairments<br />
313
(eg, limitations in bathing, feeding, or transferring), <strong>and</strong> psychosocial impairments (depression<br />
or living alone).<br />
As a patient safety practice, consultations may decrease the occurrence <strong>of</strong> iatrogenic<br />
complications such as functional decline related to hospitalization, delirium, <strong>and</strong> falls. Functional<br />
decline occurs in 25 to 60% <strong>of</strong> older persons after entering acute care, due to the interaction <strong>of</strong> a<br />
patient’s existing co-morbidities with the hospital environment. 4 It results in worsened cognitive<br />
status <strong>and</strong> physical functioning due to the stressors <strong>of</strong> hospitalization (bedrest <strong>and</strong> immobility,<br />
medical procedures <strong>and</strong> pharmacotherapy, <strong>and</strong> the hospital environment) in older patients. For<br />
information on the prevalence <strong>and</strong> severity <strong>of</strong> falls <strong>and</strong> delirium, see Chapters 26 <strong>and</strong> 28,<br />
respectively.<br />
Mortality is another important clinical outcome that may be affected. The number <strong>of</strong><br />
deaths that might be prevented by implementation <strong>of</strong> this practice is unknown, although the most<br />
common medical diagnoses <strong>of</strong> patients enrolled in these consultation studies (cardiac,<br />
pulmonary, <strong>and</strong> gastrointestinal disorders) reflect the prevalent admitting medical diagnoses for<br />
all older patients. The in-hospital, all-cause mortality in the reviewed studies (approximately 5 to<br />
15%) provides a context against which one can consider the potential for improvement if these<br />
practices influence mortality.<br />
Opportunities for Impact<br />
The number <strong>of</strong> hospitals with multidisciplinary geriatric consultation services is not<br />
reported in the literature. However, data from the American Hospital Association 5 indicate that<br />
fewer than half <strong>of</strong> hospitals <strong>of</strong>fer comprehensive geriatric assessment.* Researchers in the field<br />
believe that even in those hospitals with consultation services, only a minority <strong>of</strong> the patients<br />
most likely to benefit are being referred. Thus, if the practice is effective, there is substantial<br />
opportunity for improvement by increasing its utilization in this vulnerable patient population.<br />
Study Designs<br />
A structured literature search identified 14 controlled trials: 12 r<strong>and</strong>omized, 6-17 1<br />
alternate-allocation, 18 <strong>and</strong> 1 prospective cohort study 19 (see Table 29.1). Four <strong>of</strong> the 14 articles<br />
report different outcome measures from the same clinical trial. 6-9 One study focuses on nursing<br />
home residents; 14 all other studies were <strong>of</strong> hospitalized patients. Three <strong>of</strong> the studies were<br />
performed in Canada <strong>and</strong> one in the United Kingdom. 16-19 Two trials were limited to elderly<br />
patients with hip fractures. 15, 17 In the study by Fretwell et al, patients were admitted to a medical<br />
ward designated for seniors <strong>and</strong> staffed with specially trained nurses. 13 Because the team still<br />
functioned in a consultative role <strong>and</strong> could not implement its own recommendations, the study is<br />
included here rather in the chapter on GEM units (Chapter 30).<br />
* Of the 4953 acute medical/surgical hospitals in the American Hospital Association (AHA)<br />
database, 4398 (89%) responded to the AHA 1999 Annual Survey. Of responding hospitals,<br />
1823 (41%) indicated availability <strong>of</strong> “geriatric services,” which was defined as providing one or<br />
more <strong>of</strong> the following: comprehensive geriatric assessment, adult day care, Alzheimer’s<br />
diagnostic-assessment services, geriatric acute care units, <strong>and</strong>/or geriatric clinics. A<br />
conservative, upper-limit estimate assuming all 555 non-responding hospitals have “geriatric<br />
services” would be 48%. As the survey does not ask the availability <strong>of</strong> each type <strong>of</strong> geriatric<br />
service, the percentage <strong>of</strong> hospitals <strong>of</strong>fering inpatient comprehensive geriatric assessment based<br />
on the AHA Survey data can only be described as “less than 48%” (how much less is unknown).<br />
314
Study Outcomes<br />
Ten studies reported functional status outcomes, measured by the Katz 20 or Lawton 21<br />
index <strong>of</strong> activities <strong>of</strong> daily living, or the Barthel Index. 22 Marcantonio et al 15 measured the<br />
occurrence <strong>and</strong> severity <strong>of</strong> delirium defined according to the Confusion Assessment Method<br />
criteria. 23 Ray <strong>and</strong> colleagues reported the proportion <strong>of</strong> recurrent fallers <strong>and</strong> the rate <strong>of</strong> injurious<br />
falls in nursing home patients during one year. 14 Becker et al measured 6 classes <strong>of</strong> hospitalacquired<br />
complications: medication-related, procedures, infections, trauma or injury (eg, falls<br />
<strong>and</strong> pressure sores), psychiatric, <strong>and</strong> other (eg, urinary retention, fecal impaction). 7 Eight trials<br />
reported all-cause mortality in-hospital, at 6 months, or at one year. 8,10-13,16,18,19 A recent metaanalysis<br />
24 incorporated unpublished mortality data from several other studies 6,17 reviewed here.<br />
Other clinically relevant outcomes were changes in pharmacotherapy prescribed, length <strong>of</strong><br />
hospital stay, <strong>and</strong> discharge location.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Two non-blinded trials showed a statistically significant improvement in patients’<br />
functional ability. 17,18 Kennie et al targeted a population at high risk for functional decline during<br />
hospitalization: elderly women with hip fractures. 17 In the study by Hogan et al, 18 the difference<br />
was significant at one year but not at 3 or 6 months, suggesting that the intervention group’s<br />
post-discharge follow-up by a geriatric team may have accounted for the difference rather than<br />
prevention <strong>of</strong> iatrogenic functional decline in-hospital. The study by Thomas <strong>and</strong> colleagues 12<br />
showed a trend towards improved functional status. No other study reported improved functional<br />
outcomes.<br />
The trial by Becker et al 7 showed no significant difference in the incidence <strong>of</strong> hospitalacquired<br />
complications between intervention <strong>and</strong> control groups. Two studies that targeted<br />
specific high-risk populations did show benefit. 14,15 In the study by Marcantonio et al, a<br />
multidisciplinary consultation including assessment <strong>and</strong> targeted recommendations per a<br />
structured protocol in 10 domains (including pain treatment, bowel/bladder function, nutrition,<br />
pain treatment, mobility, <strong>and</strong> environmental stimuli) resulted in a significant decrease in<br />
perioperative delirium in patients with hip fracture. 15 Ray et al enrolled nursing home residents<br />
65+ years <strong>of</strong> age who had fallen in the previous year <strong>and</strong> had a possible problem in at least one<br />
<strong>of</strong> 4 safety domains: environmental safety, wheelchair use, psychotropic drug use, or mobility. 14<br />
<strong>Patient</strong>s who received care from the consultation team, including structured assessments <strong>and</strong><br />
specific recommendations in these safety domains, experienced a significant reduction in the rate<br />
<strong>of</strong> recurrent falls (43.8% intervention group vs. 54.1% control group, p=0.03). 14<br />
The reported r<strong>and</strong>omized clinical trials yielded mixed results for the outcome <strong>of</strong> all-cause<br />
mortality, with most studies demonstrating no benefit. The study by Thomas 12 reported a<br />
statistically significant improvement in mortality at 6 months, but Gayton et al 19 reported only a<br />
trend toward improvement at 6 months. Neither <strong>of</strong> Hogan’s studies found in-hospital mortality<br />
benefits. One study 16 showed improved mortality at 4 months <strong>and</strong> one 18 at 6 months but these<br />
benefits were not sustained at one year. Hospital-acquired complications would be expected to<br />
reduce in-hospital or short-term mortality, so the survival benefit observed many months after<br />
hospitalization in these studies suggests that other carry-over effects (eg, improved medication<br />
regimens) or better post-discharge care may be influencing these results. According to a metaanalysis<br />
24 the summary odds ratio for 6-month mortality in 8 <strong>of</strong> the studies cited 6,10,12,13,16-19 was<br />
0.77 (95% CI: 0.62-0.96), but the effect on 12-month mortality was not statistically significant.<br />
The authors tested for heterogeneity <strong>of</strong> outcomes before pooling results <strong>of</strong> the trials (p=0.07). Of<br />
315
note, the large trial (n=2353) by Reuben et al 11 was not eligible for inclusion in the meta-analysis<br />
because it was published later. Because it was larger than all other studies combined, its effect on<br />
the pooled estimate <strong>of</strong> 6-month mortality would be to reduce any statistically significant<br />
differences between intervention <strong>and</strong> study groups, since no survival advantage was reported at<br />
up to one year in the study (p=0.89 for survival curve).<br />
Potential for Harm<br />
No harm attributable to the geriatric consultation was reported in the trials.<br />
Costs <strong>and</strong> Implementation<br />
Implementation <strong>of</strong> the multidisciplinary team entails logistic planning to determine the<br />
number <strong>and</strong> type <strong>of</strong> consultative team members, <strong>and</strong> human resource coordination regarding<br />
time allocation <strong>and</strong> staffing. Few studies included data on costs <strong>of</strong> the practice, such as hospital<br />
costs incurred by assembly <strong>of</strong> the consultation team. Fretwell <strong>and</strong> colleagues, 13 however, have<br />
reported hospital charges in their study <strong>of</strong> 436 patients in a university-affiliated hospital. The<br />
percentage <strong>of</strong> patients exceeding DRG reimbursement for hospitalization was similar in both<br />
intervention <strong>and</strong> control groups, 69.7% <strong>and</strong> 71.2%, respectively. Winograd 25 reported that the<br />
cost <strong>of</strong> screening about 1200 patients to identify suitable c<strong>and</strong>idates for consultation (using<br />
predefined criteria discussed in the paper) could be accomplished by employing a trained<br />
employee working one-quarter time, at a cost (in 1998 dollars) <strong>of</strong> about $7000 over the course <strong>of</strong><br />
one year.<br />
Comment<br />
Inpatient geriatric consultation may have an impact on care for the hospitalized older<br />
patient, but the potential improvement in patient-safety outcomes is unclear. All-cause mortality<br />
differences may be due to differences in patient selection, <strong>and</strong> the data for improvement in<br />
functional outcomes suggests that certain patients may experience greater benefit than others.<br />
Appropriate targeting <strong>of</strong> services to patients at high risk for adverse outcomes such as falls <strong>and</strong><br />
delirium seems to result in benefit. Consequently, multidisciplinary geriatric consultation <strong>and</strong><br />
other efforts directed towards preventing iatrogenic functional decline, the most common<br />
complication <strong>of</strong> older hospitalized patients, deserve careful attention.<br />
Identified problems in the reviewed studies include inadequate targeting <strong>of</strong> individuals<br />
who would most benefit from the intervention, potential between-group cross-contamination,<br />
<strong>and</strong> differences in local expertise in carrying out recommended interventions. Lack <strong>of</strong><br />
effectiveness in some studies may reflect poor compliance with team suggestions or inadequate<br />
staffing to implement a consultant’s recommendations, regardless <strong>of</strong> desire to comply. Lack <strong>of</strong><br />
control over the direct management <strong>of</strong> the patient could also represent a serious shortcoming that<br />
limits effectiveness <strong>of</strong> this practice.<br />
Multidisciplinary geriatric consultative teams, in contrast to specialized geriatric<br />
evaluation <strong>and</strong> management (GEM) units, provide expertise in geriatric care throughout a<br />
hospital, but in a consultative role. In comparing this strategy with GEM units or Acute Care for<br />
Elders (ACE) units, several differences should be noted. Multidisciplinary teams are less<br />
expensive to organize <strong>and</strong> can be implemented within a shorter period <strong>of</strong> time. Since older<br />
patients reside throughout an institution, there is also greater opportunity to reach a larger<br />
number <strong>of</strong> patients when the consultation team is not single-unit based. There is no bed limit,<br />
<strong>and</strong> the capacity <strong>of</strong> the team to provide interventions is therefore limited by their available time<br />
rather than the number <strong>of</strong> beds in any one unit. The resources required to assemble an<br />
316
experienced geriatric team in a hospital that has no pre-existing geriatric expertise remains an<br />
important consideration. In addition, costs associated with enhancing <strong>and</strong> monitoring adherence<br />
with recommendations should be included when designing an effective program. Physical<br />
redesign <strong>of</strong> the unit environments to accommodate special needs (eg, special flooring, bed<br />
layout, reorienting devices) is likewise not part <strong>of</strong> this practice. Specially trained geriatric nurses<br />
are also not available equally throughout the hospital, in contrast to designated geriatric inpatient<br />
units, which have nurses focused exclusively on care <strong>of</strong> the older patient.<br />
Notwithst<strong>and</strong>ing these considerations, the practice <strong>of</strong> multidisciplinary geriatric<br />
consultation services has high face validity. More research is needed to evaluate which patients<br />
might receive maximal benefit for the associated resource commitment. Other areas for further<br />
research include examining the problems most appropriate for geriatric assessment <strong>and</strong><br />
consultation in the hospital, developing strategies to improve adherence to <strong>and</strong> execution <strong>of</strong><br />
recommendations, <strong>and</strong> identifying the components <strong>of</strong> a successful <strong>and</strong> cost-effective consultation<br />
team.<br />
317
Table 29.1. Studies <strong>of</strong> multidisciplinary geriatric consultation services<br />
Study Setting <strong>and</strong><br />
Participants<br />
Allen,<br />
1998 6<br />
Becker,<br />
1987 7<br />
Saltz, 1988 8<br />
McVey,<br />
1989 9<br />
Fretwell,<br />
1990 13<br />
Gayton,<br />
1982 19<br />
Hogan,<br />
1987 16<br />
Hogan,<br />
1990 18<br />
Kennie,<br />
1988 17<br />
185 patients at a VA<br />
hospital, 1983-1984<br />
436 patients at a<br />
university-affiliated<br />
hospital, 1985-1987<br />
222 patients at a<br />
Canadian universityaffiliated<br />
hospital, 1982-<br />
1984<br />
113 patients at a<br />
Canadian tertiary care<br />
hospital, 1984<br />
132 patients at a<br />
Canadian hospital, 1985<br />
144 orthopedic patients<br />
at a U.K. district<br />
hospital, year not stated<br />
Study<br />
Design,<br />
Outcomes<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
318<br />
Results<br />
No significant differences in hospitalacquired<br />
complications (overall 38%<br />
for both groups)<br />
No statistically significant improvement<br />
in functional status (activities <strong>of</strong> daily<br />
living)<br />
No statistically significant differences in<br />
rehospitalization or placement<br />
Compliance with recommendations:<br />
71.7% overall (from 47-95% for<br />
selected interventions)<br />
No significant difference in mortality at<br />
discharge<br />
No significant differences in length <strong>of</strong><br />
stay, physical or cognitive function, or<br />
hospital charges<br />
No significant mortality difference up to 6<br />
months follow-up, but trend favoring<br />
intervention group<br />
No significant differences in functional<br />
status, length <strong>of</strong> stay, or mental status<br />
between study groups<br />
Mortality at 4 months lower in the<br />
intervention group (p
Marcantoni<br />
o, 2001 15<br />
Ray,<br />
1997 14<br />
Reuben,<br />
1995 11<br />
Thomas,<br />
1993 12<br />
Winograd,<br />
1993 10<br />
126 orthopedic patients<br />
at an academic medical<br />
center, year not stated<br />
482 residents in 14<br />
nursing homes, 1993-<br />
1995<br />
2353 patients at 4 HMOrun<br />
hospitals, 1991-1994<br />
120 patients at a<br />
community hospital, year<br />
not stated<br />
197 men at a VA<br />
hospital, 1985-1989<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
319<br />
Occurrence <strong>of</strong> delirium: 32% vs. 50% in<br />
control group (p=0.04)<br />
Adherence to recommendations: 77%<br />
Lower rate <strong>of</strong> recurrent falls: 19% vs.<br />
54% in control group (p=0.03)<br />
Trend toward lower mean rate <strong>of</strong> injurious<br />
falls<br />
No statistically significant differences in<br />
mortality at up to one-year follow-up<br />
No significant change in functional status<br />
at 3 or 12 months<br />
Reduced 6-month mortality: 6% vs. 21%<br />
controls (p=0.01)<br />
Trend toward improved functional status<br />
in the intervention group<br />
Hospital readmission in 6-months<br />
significantly lower in the intervention<br />
group<br />
No significant mortality differences<br />
between groups<br />
No significant change in physical<br />
function, length <strong>of</strong> stay, or placement<br />
between groups<br />
Compliance with all recommendations:<br />
67%
References<br />
1. A pr<strong>of</strong>ile <strong>of</strong> older Americans: 2000. [Washington, DC]: Administration on Aging. U.S.<br />
Department <strong>of</strong> Health <strong>and</strong> Human Services, 2000.<br />
2. Kramarow E, Lentzner E, Rooks R, Weeks J, Saydah S. Health <strong>and</strong> aging chartbook.<br />
Health, United States, 1999. Hyattsville, Maryl<strong>and</strong>: National Center for Health Statistics,<br />
1999.<br />
3. Winograd CH. Targeting strategies: an overview <strong>of</strong> criteria <strong>and</strong> outcomes. J Am Geriatr<br />
Soc. 1991;39S:25S-35S.<br />
4. Palmer RM. Acute hospital care <strong>of</strong> the elderly: minimizing the risk <strong>of</strong> functional decline.<br />
Cleve Clin J Med. 1995;62:117-128.<br />
5. American Hospital Association. Health Statistics, 2001 edition. Chicago, Illinois: Health<br />
Forum, 2001.<br />
6. Allen CM, Becker PM, McVey LJ, Saltz C, Feussner JR, Cohen HJ. A r<strong>and</strong>omized,<br />
controlled clinical trial <strong>of</strong> a geriatric consultation team. Compliance with<br />
recommendations. JAMA. 1986;255:2617-2621.<br />
7. Becker PM, McVey LJ, Saltz CC, Feussner JR, Cohen HJ. Hospital-acquired complications<br />
in a r<strong>and</strong>omized controlled clinical trial <strong>of</strong> a geriatric consultation team. JAMA.<br />
1987;257:2313-2317.<br />
8. Saltz CC, McVey LJ, Becker PM, Feussner JR, Cohen HJ. Impact <strong>of</strong> a geriatric<br />
consultation team on discharge placement <strong>and</strong> repeat hospitalization. Gerontologist.<br />
1988;28:344-350.<br />
9. McVey LJ, Becker PM, Saltz CC, Feussner JR, Cohen HJ. Effect <strong>of</strong> a geriatric consultation<br />
team on functional status <strong>of</strong> elderly hospitalized patients. Ann Intern Med. 1989;110:79-84.<br />
10. Winograd CH, Gerety MB, Lai NA. A negative trial <strong>of</strong> inpatient geriatric consultation.<br />
Lessons learned <strong>and</strong> recommendations for future research. Arch Intern<br />
Med.1993;153:2017-2023.<br />
11. Reuben DB, Borok GM, Wolde-Tsadik G, et al. A r<strong>and</strong>omized trial <strong>of</strong> comprehensive<br />
geriatric assessment in the care <strong>of</strong> hospitalized patients. N Engl J Med. 1995;332:1345-<br />
1350.<br />
12. Thomas DR, Brahan R, Haywood BP. Inpatient community-based geriatric assessment<br />
reduces subsequent mortality. J Am Geriatr Soc. 1993;41:101-104.<br />
13. Fretwell MD, Raymond PM, McGarvey ST, et al. The Senior Care Study. A controlled trial<br />
<strong>of</strong> a consultative/unit-based geriatric assessment program in acute care. J Am Geriatr Soc.<br />
1990;38:1073-1081.<br />
14. Ray WA, Taylor JA, Meador KG, et al. A r<strong>and</strong>omized trial <strong>of</strong> a consultation service to<br />
reduce falls in nursing homes. JAMA. 1997;278:557-562.<br />
15. Marcantonio ER, Flacker JM, Wright RJ, Resnick NM. Reducing delirium after hip<br />
fracture: a r<strong>and</strong>omized trial. J Am Geriatr Soc. 2001;49:In press.<br />
16. Hogan DB, Fox RA, Badley BW, Mann OE. Effect <strong>of</strong> a geriatric consultation service on<br />
management <strong>of</strong> patients in an acute care hospital. CMAJ. 1987;136:713-717.<br />
17. Kennie DC, Ried J, Richardson IR, Kiamari AA, Kelt C. Effectiveness <strong>of</strong> geriatric<br />
rehabilitative care after fractures <strong>of</strong> the proximal femur in elderly women: a r<strong>and</strong>omised<br />
clinical trial. BMJ. 1988;297:1083-1086.<br />
18. Hogan DB, Fox RA. A prospective controlled trial <strong>of</strong> a geriatric consultation team in an<br />
acute-care hospital. Age Ageing.1990;19:107-113.<br />
320
19. Gayton D, Wood-Dauphinee S, de Lorimer M, Tousignant P, Hanley J. Trial <strong>of</strong> a geriatric<br />
consultation team in an acute care hospital. J Am Geriatr Soc. 1987;35:726-736.<br />
20. Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies <strong>of</strong> illness in the aged.<br />
The index <strong>of</strong> ADL: a st<strong>and</strong>ardized measure <strong>of</strong> biological <strong>and</strong> psychosocial function. JAMA.<br />
1963;185:914-919.<br />
21. Lawton MP, Brody EM. Assessment <strong>of</strong> older people: self-maintaining <strong>and</strong> instrumental<br />
activities <strong>of</strong> daily living. Gerontologist. 1969;9:179-186.<br />
22. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J.<br />
1965;14:61-65.<br />
23. Inouye SK, Van Dycke CH, Alessi CA, Balkin S, Siegal AP, Horwitz RI. Clarifying<br />
confusion: the Confusion Assessment Method. Ann Intern Med. 1990;113:941-948.<br />
24. Stuck AE, Siu AL, Wiel<strong>and</strong> GD, Adams J, Rubenstein LZ. Comprehensive geriatric<br />
assessment: a meta-analysis <strong>of</strong> controlled trials. Lancet. 1993;342:1032-1036.<br />
25. Winograd CH, Gerety MB, Brown E, Kolodny V. Targeting the hospitalized elderly for<br />
geriatric consultation. J Am Geriatr Soc. 1988;36:1113-1119.<br />
321
322
Chapter 30. Geriatric Evaluation <strong>and</strong> Management Units for Hospitalized<br />
<strong>Patient</strong>s<br />
Joseph V. Agostini, MD<br />
Dorothy I. Baker, PhD, RNCS<br />
Sidney T. Bogardus, Jr., MD<br />
Yale University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Inpatient care <strong>of</strong> elders with multiple co-morbidities requires close attention to the special<br />
needs <strong>and</strong> geriatric syndromes that arise in this vulnerable population. 1,2 One strategy to address<br />
the risks <strong>of</strong> hospitalization is to provide care by a multidisciplinary team in a dedicated geriatric<br />
unit, principally Geriatric Evaluation <strong>and</strong> Management (GEM) Units. This model <strong>of</strong> care, in<br />
many ways similar to coordinated stroke units, 3 may improve mortality <strong>and</strong> other clinically<br />
relevant outcomes compared with outcomes achieved on a general medical ward. An alternative<br />
strategy, reviewed elsewhere in this Report, is the comprehensive geriatric consultation service,<br />
analogous to a typical medical consultation team (see Chapter 29).<br />
In a GEM unit, a multidisciplinary team provides comprehensive geriatric assessment,<br />
detailed treatment plans, <strong>and</strong> attention to the rehabilitative needs <strong>of</strong> older patients. A typical team<br />
is composed <strong>of</strong> a geriatrician, clinical nurse specialist, social worker, <strong>and</strong> specialists from such<br />
fields as occupational <strong>and</strong> physical therapy, nutrition, pharmacy, audiology, <strong>and</strong> psychology.<br />
GEM units are typically separate hospital wards that have been redesigned to facilitate care <strong>of</strong><br />
the geriatric patient. Multidisciplinary team rounds <strong>and</strong> patient-centered team conferences are<br />
hallmarks <strong>of</strong> care on these units, which, in contrast to geriatric consultation services, have direct<br />
control over the implementation <strong>of</strong> team recommendations.<br />
Practice Description<br />
In all the studies reviewed in this chapter, the GEM unit team included a physician<br />
experienced in geriatric medicine, skilled geriatric nurses <strong>and</strong> rehabilitation specialists. (The<br />
latter may not have been on site but were accessible). The teams completed multidimensional<br />
geriatric assessments, conducted interdisciplinary team rounds, <strong>and</strong> provided comprehensive<br />
discharge planning. Units were physically separate wards that were designed to facilitate<br />
geriatric care. Acute Care <strong>of</strong> Elders (ACE) Units 4 incorporate the GEM unit design with<br />
additional enhancements <strong>and</strong> admit patients with acute illnesses. ACE units <strong>of</strong>ten provide<br />
improved flooring <strong>and</strong> lighting, reorienting devices, <strong>and</strong> other environmental improvements such<br />
as common rooms for patient use. For example, in the ACE unit studied by L<strong>and</strong>efeld <strong>and</strong><br />
colleagues 5 the GEM unit concept was enhanced by the use <strong>of</strong> more nurse-initiated protocols <strong>and</strong><br />
a greater number <strong>of</strong> environmental <strong>and</strong> design modifications.<br />
Both styles <strong>of</strong> unit emphasize the early assessment <strong>of</strong> risk factors for iatrogenic<br />
complications <strong>and</strong> the prevention <strong>of</strong> functional decline. 4,6 Studies <strong>of</strong> both are included in this<br />
chapter.<br />
323
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
One-third <strong>of</strong> hospitalized patients are aged 65 years <strong>and</strong> older. In 1996, although<br />
comprising only 13% <strong>of</strong> the US population they accounted for 38% <strong>of</strong> the approximately 31<br />
million discharges from non-government, acute care hospitals. 7 Since all hospitalized older<br />
patients are potentially at risk, the target population is quite large. Appropriate selection <strong>of</strong><br />
patients who are at risk for hospital-related complications <strong>and</strong> who are likely to receive benefit<br />
from this practice, however, would decrease this number. 8<br />
The target safety problems are preventable deaths <strong>and</strong> hospital-related functional decline<br />
in older persons. The number <strong>of</strong> deaths that could be prevented if the practice were to be widely<br />
implemented is unknown. On the other h<strong>and</strong>, since hospital-related functional decline occurs in<br />
25 to 60% <strong>of</strong> older hospitalized patients, there is substantial opportunity to improve clinical<br />
outcomes. 9 Other clinical problems explicitly studied in controlled trials include cognitive status<br />
<strong>and</strong> nursing home placement.<br />
Opportunities for Impact<br />
Data from the American Hospital Association (AHA) 10 indicate that fewer than half <strong>of</strong><br />
hospitals providing care for the elderly have geriatric acute care units or <strong>of</strong>fer comprehensive<br />
geriatric assessment. (see footnote in Chapter 29). Researchers in the field believe the number is<br />
increasing. The Department <strong>of</strong> Veterans Affairs reports that in 1997 there were 110 active GEM<br />
units, although some concentrate solely on outpatient assessment. 11 A recent national survey 12<br />
identified at least 15 active ACE units, with average daily censuses ranging from 5 to 25<br />
patients. Depending on the screening <strong>and</strong> targeting criteria used to identify eligible patients, the<br />
potential for impact could be quite large <strong>and</strong> raises the question <strong>of</strong> the physical <strong>and</strong> manpower<br />
capacity <strong>of</strong> these units to meet the apparent dem<strong>and</strong>.<br />
Study Designs<br />
A structured literature search identified a systematic review 13 that included 6 studies (4<br />
r<strong>and</strong>omized controlled trials, 14-17 one retrospective cohort study with historical controls, 18 <strong>and</strong><br />
one published abstract <strong>of</strong> a r<strong>and</strong>omized controlled trial 19 ). We also identified 2 r<strong>and</strong>omized<br />
controlled trials <strong>of</strong> ACE units 5,20 that were published later (see Table 30.1). All <strong>of</strong> the cited<br />
studies were single center in design. Five <strong>of</strong> the studies 5,14,16,17,20 provided sufficient data to<br />
evaluate the baseline level <strong>of</strong> function <strong>of</strong> enrolled patients.<br />
Study Outcomes<br />
All-cause mortality was reported in each study. For this outcome, most patients were<br />
followed 6 or more months after hospitalization. Other clinically important outcomes measured<br />
in some studies were functional status, 5,14,16,17,20 cognitive status, 14,16,17 length <strong>of</strong> stay, 5,14-16,18,20<br />
<strong>and</strong> discharge rates to institutional settings. 5,14-20<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
In some studies mortality during hospitalization, at 3 months, or at 6 months was reduced<br />
in the intervention group but the differences failed to achieve statistical significance. A metaanalysis<br />
13 <strong>of</strong> 6 studies 14-19 found the summary odds ratio for 6-month mortality was 0.65 (95%<br />
CI: 0.46-0.91), using both published <strong>and</strong> unpublished data from the included trials. Tests for<br />
heterogeneity across studies were reported with p
Cognitive function, as measured by the Kahn-Goldfarb Mental Status Questionnaire, 21<br />
showed no statistical improvement over the course <strong>of</strong> one study, 17 nor did 2 other studies 14,16<br />
demonstrate improvement, using the Mini-Mental State Examination 22 to assess mental status.<br />
Two trials <strong>of</strong> ACE units examined functional status, using the basic activities <strong>of</strong> daily living<br />
(ADL). 23 L<strong>and</strong>efeld et al 5 reported statistically significant improvements, while Counsell et al 20<br />
found benefit in a composite outcome <strong>of</strong> ADL improvement <strong>and</strong> nursing home placement, but<br />
not in discharge ADL levels alone. Two other studies also demonstrated statistically improved<br />
functional status in the six months after r<strong>and</strong>omization 14 <strong>and</strong> at 12 months follow-up. 17<br />
In individual studies, GEM units were associated with a higher likelihood <strong>of</strong> home<br />
residence, rather than in an institutional setting (skilled nursing facility or nursing home).<br />
325<br />
5,14, 17<br />
The meta-analysis by Stuck et al calculated a combined odds ratio that revealed a statistically<br />
significant increase in patients discharged from GEM units who were living at home at 6 months<br />
(summary odds ratio 1.80, 95% CI: 1.28-2.53) <strong>and</strong> 12 months (summary odds ratio 1.68, 95%<br />
CI: 1.17-2.41) thereafter. 13 ACE unit trials in a community hospital 20 <strong>and</strong> a university hospital 5<br />
were also both successful in decreasing patient placement in institutional settings <strong>and</strong> would<br />
likely have strengthened the summary estimate if included. Extrapolating their study findings to<br />
the US population, Rubenstein <strong>and</strong> colleagues 17 estimated that approximately 200,000 nursing<br />
home admissions per year could be avoided using their geriatric evaluation unit approach.<br />
Winograd noted that this multidisciplinary practice would potentially be more effective if target<br />
populations were better identified <strong>and</strong> enrolled using specific criteria. 8<br />
Potential for Harm<br />
No data suggest that GEM units were associated with harm.<br />
Costs <strong>and</strong> Implementation<br />
Implementation <strong>of</strong> the practice requires construction or redesign <strong>of</strong> hospital ward(s) to<br />
create a suitable environment, training or recruiting experienced staff, establishing selection<br />
criteria to determine patient eligibility, <strong>and</strong> implementing a continuous evaluation process to<br />
assess clinical <strong>and</strong> non-clinical outcomes.<br />
A working group has recommended including costs as an important outcome measure in<br />
future studies <strong>of</strong> GEM units. 24 Applegate <strong>and</strong> colleagues reported in a later analysis 25 <strong>of</strong> their<br />
r<strong>and</strong>omized controlled trial 14 that the increased costs associated with their intervention study<br />
were not balanced by savings in subsequent health care spending, but if charges were adjusted<br />
for days spent at home (versus in long-term care) the charges were similar. Lower direct costs<br />
were demonstrated by one intervention study, 17 particularly after adjusting for differences in<br />
survival (mean institutional-care costs per year survived, $22,597 for intervention patients vs.<br />
$27,826 for control-group patients). The study by L<strong>and</strong>efeld et al 5 reported mean hospital<br />
charges <strong>of</strong> $10,289 for intervention patients compared with $12,412 for control patients, with<br />
similar median charges (p=0.3). Additional costs <strong>of</strong> about $75,000 attributable to staffing <strong>and</strong><br />
physical redesign <strong>of</strong> the unit resulted in a cost <strong>of</strong> about $230 per patient in the intervention<br />
group. Counsell et al, 20 in a recent large r<strong>and</strong>omized trial <strong>of</strong> an ACE intervention, reported no<br />
difference in hospital costs for patients in the intervention group compared with the usual care<br />
group ($5640 vs. $5754, respectively; p=0.68). Included in the costs was $28 per hospital day<br />
per intervention patient, representing costs <strong>of</strong> the geriatric nurse specialist, unit medical director,<br />
<strong>and</strong> unit renovations ($300,000).
Comment<br />
Reasonable evidence supports the use <strong>of</strong> GEM units, despite varying findings across<br />
individual studies with respect to their effectiveness at preventing outcomes such as mortality.<br />
Nonetheless, mortality appears to be improved after pooling results <strong>of</strong> smaller trials. There is<br />
good evidence that this model <strong>of</strong> care decreases nursing home placements, which in itself is a<br />
noteworthy finding. Furthermore, the intervention costs may not be significantly higher in these<br />
specialized units. The generalizability <strong>of</strong> the practice requires further examination, <strong>and</strong> the need<br />
for multicenter studies, as advocated by a previous consensus group, 26 has thus far not been<br />
undertaken.<br />
Limitations <strong>of</strong> this practice compared with multidisciplinary geriatric consultation teams<br />
(Chapter 29) include limited bed availability in most units, decreased transferability <strong>of</strong> geriatric<br />
practices throughout a hospital, <strong>and</strong> a larger resource commitment compared with a hospitalwide<br />
consultation team. The advantages <strong>of</strong> the GEM <strong>and</strong> ACE unit model are direct control over<br />
implementation <strong>of</strong> clinical recommendations, the presence <strong>of</strong> dedicated geriatric nursing <strong>and</strong><br />
rehabilitative staff associated with the unit, <strong>and</strong> the beneficial effects <strong>of</strong> a ward designed to<br />
address older patients’ needs. In sum, the practice <strong>of</strong> a dedicated GEM or ACE unit carries much<br />
promise.<br />
326
Table 30.1. R<strong>and</strong>omized Controlled Trials <strong>of</strong> Geriatric Evaluation <strong>and</strong> Management Units*<br />
Study Study Setting Study<br />
Design<br />
Stuck,<br />
1993 13<br />
Applegat<br />
e, 1990 14<br />
Counsell,<br />
2000 20<br />
Gilchrist,<br />
1998 15<br />
Harris,<br />
1991 16<br />
L<strong>and</strong>efel<br />
d, 1995 5<br />
Powell,<br />
1990 19<br />
6 studies (published 1983-<br />
1991) in the US, UK,<br />
Australia, <strong>and</strong> Canada,<br />
involving 1090 patients<br />
(meta-analysis <strong>of</strong> Refs. 14-<br />
19)<br />
155 patients in a<br />
university-affiliated<br />
community hospital, 1985-<br />
1987<br />
1531 patients in a<br />
community hospital, 1994-<br />
1997<br />
222 women on an<br />
orthopedic-geriatric service<br />
in the UK, 1984-1986<br />
267 patients in an<br />
Australian hospital, 1985-<br />
1986<br />
651 patients in a<br />
university-affiliated<br />
hospital, 1990-1992<br />
203 patients in two<br />
Canadian teaching<br />
hospitals, year not stated<br />
327<br />
All-Cause Mortality <strong>and</strong> Other Outcomes†<br />
Level 1A 6-month mortality: summary odds ratio 0.65<br />
(95% CI: 0.46-0.91);<br />
12-month mortality: summary odds ratio<br />
0.77 (95% CI: 0.56-1.06)<br />
Level 1 6-month mortality: 10% vs. 21% (p=NS);<br />
After 6 months: greatest difference p=0.08<br />
by log-rank test;<br />
Improvement in ADLs: 3 <strong>of</strong> 8 ADLs better<br />
in intervention group (p0.1)<br />
Level 1 Inpatient mortality: estimated from Figure 1<br />
in paper, 8% vs. 6% (p=NS);<br />
12-month mortality: 23% vs. 29% (p=NS)<br />
Level 1 Inpatient mortality: 7% in both groups<br />
(p=NS);<br />
3-month mortality: 14% vs. 13% (p=NS);<br />
Improvement in ADLs at discharge: 34%<br />
vs. 24% (p=0.009);<br />
Discharged to nursing home: 14% vs. 22%<br />
(p=0.01)<br />
Level 1 Mortality: lower in intervention group;<br />
timing not stated (p not stated);<br />
Transferred to long-term care: fewer in<br />
intervention group (p not stated)
Rubenstei<br />
n, 1984 17<br />
Teasdale,<br />
1983 18<br />
123 patients in a VA<br />
hospital, 1980-1982<br />
124 patients in a VA<br />
hospital, 1981-1982<br />
Level 1 Inpatient mortality: 14.3% vs. 15.0%<br />
(p=NS);<br />
12-month mortality: 23.8% vs. 48.3%<br />
(p
References<br />
1. Creditor MC. Hazards <strong>of</strong> hospitalization <strong>of</strong> the elderly. Ann Intern Med. 1993;118:219-223.<br />
2. Hirsch CH, Sommers L, Olsen A, Mullen L, Winograd CH. The natural history <strong>of</strong><br />
functional morbidity in hospitalized older patients. J Am Geriatr Soc. 1990;38:1296-1303.<br />
3. Stroke Unit Trialists Collaboration. How do stroke units improve patient outcomes? A<br />
collaborative systematic review <strong>of</strong> the r<strong>and</strong>omized trials. Stroke. 1997;28:2139-2144.<br />
4. Palmer RM, Counsell S, L<strong>and</strong>efeld CS. Clinical intervention trials: the ACE unit. Clin<br />
Geriatr Med. 1998;14:831-849.<br />
5. L<strong>and</strong>efeld CS, Palmer RM, Kresevic DM, Fortinsky RH, Kowal J. A r<strong>and</strong>omized trial <strong>of</strong><br />
care in a hospital medical unit especially designed to improve the functional outcomes <strong>of</strong><br />
acutely ill older patients. N Engl J Med.1995;332:1338-1344.<br />
6. Palmer RM, L<strong>and</strong>efeld CS, Kresevic D, Kowal J. A medical unit for the acute care <strong>of</strong> the<br />
elderly. J Am Geriatr Soc. 1994;42:545-552.<br />
7. Kramarow E, Lentzner E, Rooks R, Weeks J, Saydah S. Health <strong>and</strong> aging chartbook.<br />
Health, United States, 1999. Hyattsville, Maryl<strong>and</strong>: National Center for Health Statistics,<br />
1999.<br />
8. Winograd CH. Targeting strategies: an overview <strong>of</strong> criteria <strong>and</strong> outcomes. J Am Geriatr<br />
Soc. 1991;39S:25S-35S.<br />
9. Palmer RM. Acute hospital care <strong>of</strong> the elderly: minimizing the risk <strong>of</strong> functional decline.<br />
Cleve Clin J Med.1995;62:117-128.<br />
10. American Hospital Association. Health Statistics, 2001 edition. Chicago, Illinois: Health<br />
Forum, 2001.<br />
11. Veterans Administration. VA programs for senior veterans. Geriatric evaluation <strong>and</strong><br />
management program. Available at: http://www.va.gov/seniors/health/gem.htm. Accessed<br />
May 14, 2001.<br />
12. Raziano DB, Jayadevappa R, Valenzula D, Weiner M, Lavizzo-Mourey RJ. Do Acute Care<br />
for the Elderly (ACE) units work? Paper presented at: American Geriatrics Society Annual<br />
Meeting; May 10, 2001; Chicago, IL. 2001.<br />
13. Stuck AE, Siu AL, Wiel<strong>and</strong> GD, Adams J, Rubenstein LZ. Comprehensive geriatric<br />
assessment: a meta-analysis <strong>of</strong> controlled trials. Lancet. 1993;342:1032-1036.<br />
14. Applegate WB, Miller ST, Graney MJ, Elam JT, Burns R, Akins DE. A r<strong>and</strong>omized,<br />
controlled trial <strong>of</strong> a geriatric assessment unit in a community rehabilitation hospital. N Engl<br />
J Med. 1990;322:1572-1578.<br />
15. Gilchrist WJ, Newman RJ, Hamblen DL, Williams BO. Prospective r<strong>and</strong>omised study <strong>of</strong><br />
an orthopaedic geriatric inpatient service. BMJ. 1988;297:1116-1118.<br />
16. Harris RD, Henschke PJ, Popplewell PY, et al. A r<strong>and</strong>omised study <strong>of</strong> outcomes in a<br />
defined group <strong>of</strong> acutely ill elderly patients managed in a geriatric assessment unit or a<br />
general medical unit. Aust N Z J Med. 1991;21:230-234.<br />
17. Rubenstein LZ, Josephson KR, Wiel<strong>and</strong> GD, English PA, Sayre JA, Kane RL.<br />
Effectiveness <strong>of</strong> a geriatric evaluation unit. A r<strong>and</strong>omized clinical trial. N Engl J Med.<br />
1984;311:1664-1670.<br />
18. Teasdale TA, Shuman L, Snow E, Luchi RJ. A comparison <strong>of</strong> placement outcomes <strong>of</strong><br />
geriatric cohorts receiving care in a geriatric assessment unit <strong>and</strong> on general medicine<br />
floors. J Am Geriatr Soc. 1983;31:529-534.<br />
19. Powell C, Montgomery P. The age study: the admission <strong>of</strong> geriatric patients through<br />
emergency [abstract]. Age & Ageing. 1990;19(Suppl. 2):P21-P22.<br />
329
20. Counsell SR, Holder CM, Liebenauer MA, et al. Effects <strong>of</strong> a multicomponent intervention<br />
on functional outcomes <strong>and</strong> process <strong>of</strong> care in hospitalized older patients: a r<strong>and</strong>omized<br />
controlled trial <strong>of</strong> Acute Care for Elders (ACE) in a community hospital. J Am Geriatr Soc.<br />
2000;48:1572-1581.<br />
21. Kahn RL, Goldfarb AI, Pollack M, Peck A. Brief objective measures for the determination<br />
<strong>of</strong> mental states in the aged. Am J Psychiatry. 1960;117:326-328.<br />
22. Folstein MF, Folstein SE, McHugh PR. "Mini-mental state": a practical method for grading<br />
the cognitive state <strong>of</strong> patients by the clinician. J Psychiatr Res. 1975;12:189-198.<br />
23. Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies <strong>of</strong> illness in the aged.<br />
The index <strong>of</strong> ADL: a st<strong>and</strong>ardized measure <strong>of</strong> biological <strong>and</strong> psychosocial function. JAMA.<br />
1963;185:914-919.<br />
24. Hedrick SC, Barr<strong>and</strong> N, Deyo R, et al. Working group recommendations: measuring<br />
outcomes <strong>of</strong> care in geriatric evaluation <strong>and</strong> management units. J Am Geriatr Soc.<br />
1991;39S:48S-52S.<br />
25. Applegate WB, Graney MJ, Miller ST, Elam JT. Impact <strong>of</strong> a geriatric assessment unit on<br />
subsequent health care charges. Am J <strong>Public</strong> Health. 1991;81:1302-1306.<br />
26. Kramer A, Deyo R, Applegate W, Meehan S. Research strategies for geriatric evaluation<br />
<strong>and</strong> management: conference summary <strong>and</strong> recommendations. J Am Geriatr Soc.<br />
1991;39S:53S-57S.<br />
330
Section E. General Clinical Topics<br />
Chapter 31. Prevention <strong>of</strong> Venous Thromboembolism<br />
Chapter 32. Prevention <strong>of</strong> Contrast-Induced Nephropathy<br />
Chapter 33. Nutritional Support<br />
Chapter 34. Prevention <strong>of</strong> Clinically Significant Gastrointestinal Bleeding in Intensive Care<br />
Unit <strong>Patient</strong>s<br />
Chapter 35. Reducing Errors in the Interpretation <strong>of</strong> Plain Radiographs <strong>and</strong> Computed<br />
Tomography Scans<br />
Chapter 36. Pneumococcal Vaccination Prior to Hospital Discharge<br />
Chapter 37. Pain Management<br />
Subchapter 37.1. Use <strong>of</strong> Analgesics in the Acute Abdomen<br />
Subchapter 37.2. Acute Pain Services<br />
Subchapter 37.3. Prophylactic Antiemetics During <strong>Patient</strong>-controlled Analgesia<br />
Therapy<br />
Subchapter 37.4. Non-pharmacologic Interventions for Postoperative Plan<br />
331
332
Chapter 31. Prevention <strong>of</strong> Venous Thromboembolism<br />
Jennifer Kleinbart, MD<br />
Mark V. Williams, MD<br />
Kimberly Rask, MD, PhD<br />
Emory University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Venous thromboembolism (VTE) refers to occlusion within the venous system. It<br />
includes deep vein thrombosis (DVT), typically <strong>of</strong> the lower extremities, <strong>and</strong> embolism to the<br />
pulmonary vasculature. Distal DVT is occlusion confined to the deep calf veins, while<br />
thrombosis at or above the popliteal vein is considered proximal DVT. A distal DVT becomes<br />
clinically important if it extends proximally, where the chance <strong>of</strong> pulmonary embolization (PE)<br />
is clinically significant.<br />
In hospitalized patients, VTE occurs with relatively high frequency. A patient’s risk <strong>of</strong><br />
VTE varies depending on multiple factors including age, medical condition, type <strong>of</strong> surgery,<br />
duration <strong>of</strong> immobilization, <strong>and</strong> the presence <strong>of</strong> an underlying hypercoagulable state such as<br />
malignancy. Measures to prevent VTE have been widely studied for many reasons, including<br />
VTE’s high incidence, its associated mortality <strong>and</strong> morbidity, its cost <strong>of</strong> treatment, <strong>and</strong> its<br />
treatment-related complications.<br />
VTE is <strong>of</strong>ten clinically silent. As a result, studies evaluating the efficacy <strong>of</strong> preventive<br />
measures generally screen patients who are asymptomatic. As widespread screening is not<br />
recommended in general practice, the incidence <strong>of</strong> VTE in most studies appears higher than that<br />
encountered in clinical practice. The importance <strong>of</strong> clinically undetected VTE is not fully<br />
understood.<br />
Practice Description<br />
Both mechanical <strong>and</strong> pharmacologic interventions have been evaluated for prevention <strong>of</strong><br />
VTE (Table 31.1). Mechanical devices include graduated elastic stockings (ES) <strong>and</strong> intermittent<br />
pneumatic compression (IPC); pharmacologic measures include low dose unfractionated heparin<br />
(LDUH), low molecular weight heparin (LMWH), warfarin, <strong>and</strong> aspirin .<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problems<br />
There are over 23 million surgeries performed each year in the United States. 1 The<br />
frequency <strong>of</strong> DVT <strong>and</strong> PE varies by type <strong>of</strong> procedure <strong>and</strong> specific patient risk factors, as set<br />
forth below. In general, without prophylaxis DVT occurs after approximately 20% <strong>of</strong> all major<br />
surgical procedures <strong>and</strong> PE in 1-2%. Over 50% <strong>of</strong> major orthopedic procedures are complicated<br />
by DVT <strong>and</strong> up to 30% by PE if prophylactic treatment is not instituted. 2<br />
In addition, <strong>of</strong> the more than 31 million patients admitted each year for medical<br />
conditions, up to 16% <strong>of</strong> patients will develop DVT in the absence <strong>of</strong> prophylaxis. 1,2<br />
333
Opportunities for Impact<br />
Despite the frequency with which VTE occurs in hospitalized patients, <strong>and</strong> the wellestablished<br />
efficacy <strong>and</strong> safety <strong>of</strong> preventative measures, prophylaxis is <strong>of</strong>ten underused or used<br />
inappropriately. One survey <strong>of</strong> general surgeons found that 14% did not use VTE prophylaxis. 3<br />
Another survey <strong>of</strong> orthopedic surgeons found that only 55% placed all hip fracture patients on<br />
VTE prophylaxis, <strong>and</strong> 12% never used prophylaxis. 4 When performing total hip replacement<br />
(THR) <strong>and</strong> total knee replacement (TKR), 81-84% <strong>of</strong> surgeons placed all patients on<br />
prophylaxis, <strong>and</strong> 3-5% used no prophylaxis. Taken together, these data imply that VTE<br />
prophylaxis is used in 92% <strong>of</strong> THR patients, 89% <strong>of</strong> TKR patients, <strong>and</strong> 73% <strong>of</strong> hip fracture<br />
repairs. For spinal procedures, 21-38% <strong>of</strong> surgeons used prophylaxis with all patients, while 46-<br />
64% did not use prophylaxis with any patients, an overall rate <strong>of</strong> prophylaxis <strong>of</strong> 25-44%. 4 A<br />
more recent chart review <strong>of</strong> Medicare patients over age 65 undergoing major abdominothoracic<br />
surgery from 20 Oklahoma hospitals found that only 38% <strong>of</strong> patients were given VTE<br />
prophylaxis. 5 Of patients considered at very high risk for VTE, the same percentage received<br />
some form <strong>of</strong> prophylaxis, but only 66% <strong>of</strong> those received appropriate preventive measures. 5<br />
Another chart review <strong>of</strong> high-risk patients at 10 acute care US hospitals found that some<br />
type <strong>of</strong> VTE prophylaxis was used in 94% <strong>of</strong> orthopedic patients <strong>and</strong> 75% <strong>of</strong> abdominal surgery<br />
patients. 6 However, the 1995 American College <strong>of</strong> Chest Physician grade A recommendations 7<br />
(those therapies supported by the most rigorously designed studies) were followed in only 84%<br />
<strong>of</strong> THRs, 76% <strong>of</strong> TKRs, 45% <strong>of</strong> hip fracture repairs, <strong>and</strong> 50% <strong>of</strong> abdominal surgery patients.<br />
Compliance with grade A recommendations was no better in the 3 hospitals that had critical<br />
pathways, pocket guides, or a policy for VTE prophylaxis. 6<br />
Inadequate VTE prophylaxis among medical patients may be even more prevalent. In one<br />
study, only 33% <strong>of</strong> patients admitted to a medical intensive care unit received VTE prophylaxis.<br />
Eighty-seven percent <strong>of</strong> these patients had one risk factor, <strong>and</strong> over 50% had multiple risks for<br />
VTE. 8 Another study found that <strong>of</strong> patients who developed VTE during admission or within 30<br />
days <strong>of</strong> discharge, 48% had not received prophylaxis during hospitalization. Most <strong>of</strong> these were<br />
medical patients. 9<br />
Study Designs <strong>and</strong> Outcomes<br />
There are a large number <strong>of</strong> r<strong>and</strong>omized control trials (RCTs) <strong>and</strong> high quality metaanalyses<br />
that examine the efficacy <strong>of</strong> VTE prophylaxis. Most studies considered surgical rather<br />
than medical patients. Several differences in study design may account for heterogeneity among<br />
these studies, including differences in patients or procedures, the intervention (type, duration, or<br />
dose <strong>of</strong> prophylaxis), the method used to diagnosis VTE, the outcome measured (distal or<br />
proximal DVT, fatal or nonfatal PE), <strong>and</strong> whether the endpoint was a clinical event or one found<br />
by routine screening.<br />
The “gold st<strong>and</strong>ard” for diagnosis <strong>of</strong> DVT is contrast venography, which is generally<br />
used in studies screening high-risk patients. Venography may detect a significant number <strong>of</strong> clots<br />
that are not clinically important, <strong>and</strong> is <strong>of</strong>ten technically limited. 10 Other common screening<br />
methods are fibrinogen leg scanning (fibrinogen uptake test, FUT) <strong>and</strong> duplex ultrasonography,<br />
both <strong>of</strong> which have low sensitivity for calf vein thrombosis. 11,12 Meta-analysis <strong>of</strong> studies using<br />
the FUT in orthopedic patients showed a sensitivity <strong>of</strong> 55% for calf vein thrombosis <strong>and</strong> 45% for<br />
all DVT, with a specificity <strong>of</strong> 92%. 11 For detection <strong>of</strong> calf vein thrombosis, duplex<br />
ultrasonography had a sensitivity <strong>and</strong> specificity <strong>of</strong> 39% <strong>and</strong> 98% in symptomatic hospitalized<br />
patients, <strong>and</strong> <strong>of</strong> 13% <strong>and</strong> 92% in patients undergoing arthroplasty. 12 In studies <strong>of</strong> general surgery<br />
334
patients, the incidence <strong>of</strong> DVT was 25% when diagnosis was made by FUT, <strong>and</strong> 19% when<br />
confirmed by venogram. 2 However, in trials <strong>of</strong> VTE prophylaxis using elastic stockings in<br />
neurosurgical patients, the study using FUT found a DVT rate <strong>of</strong> 9%, while 3 others using<br />
venography found the rate <strong>of</strong> DVT to be 28%. 2<br />
In many studies, all patients are screened for DVT, although several studies only tested<br />
for DVT if clinical signs or symptoms occurred. Clearly, in the latter a number <strong>of</strong> DVT are<br />
missed; however, the clinical importance <strong>of</strong> these asymptomatic clots is unknown.<br />
The gold st<strong>and</strong>ard for diagnosis <strong>of</strong> pulmonary embolus is pulmonary angiography.<br />
However, reliable evidence confirms that combining clinical probability with results <strong>of</strong><br />
ventilation-perfusion scanning is an accurate <strong>and</strong> less invasive method <strong>of</strong> making this<br />
diagnosis. 13 Most studies report the incidence <strong>of</strong> symptomatic or fatal PE. Studies rarely screen<br />
for PE in asymptomatic patients. Silent pulmonary embolism may occur in up to 50% <strong>of</strong> patients<br />
with proximal DVT, however, the clinical importance <strong>of</strong> these emboli is unclear. 14<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Studies <strong>of</strong> VTE prophylaxis are best grouped by the population at risk (Table 31.2). The<br />
sections that follow describe the prevalence <strong>of</strong> thromboembolism in patients in each category<br />
<strong>and</strong> discuss the efficacy <strong>of</strong> various prophylactic strategies.<br />
General Surgery<br />
For general surgery patients not receiving VTE prophylaxis, the incidence <strong>of</strong> DVT<br />
confirmed by venogram is 19%. Proximal DVT occurs in 7% <strong>of</strong> patients, PE in 1.6%, <strong>and</strong> fatal<br />
PE in 0.9%. 2<br />
In general surgery patients, the risk <strong>of</strong> VTE is highest in those undergoing major surgery<br />
who are over the age <strong>of</strong> 40 <strong>and</strong> have an underlying hypercoagulable state, prior VTE, or cancer.<br />
(<strong>Patient</strong>s with spinal cord injury, trauma, <strong>and</strong> those undergoing hip or knee arthroplasty or hip<br />
fracture surgery also fall in this risk group <strong>and</strong> are discussed separately below). Incidence rates<br />
for patients in this “very high” risk group are as follows: calf DVT, up to 80%; proximal DVT,<br />
20%; PE, 10%; <strong>and</strong> up to 5% may suffer a fatal PE if no prophylaxis is provided. 2 The risk <strong>of</strong><br />
VTE is lowest in those surgical patients who are young, otherwise healthy, <strong>and</strong> whose surgeries<br />
are “minor.” Such patients have incident rates as follows: calf DVT, 2%; proximal DVT, 0.4%,<br />
PE , 0.2%; <strong>and</strong> fatal PE , 0.002%.<br />
Numerous studies show that both LDUH <strong>and</strong> LMWH reduce the risk <strong>of</strong> proximal DVT,<br />
PE, <strong>and</strong> fatal PE in patients undergoing general surgery. Pooled results from 46 r<strong>and</strong>omized<br />
trials show that prophylaxis <strong>of</strong> general surgical patients with LDUH compared with placebo<br />
reduced the risk <strong>of</strong> DVT (diagnosed by FUT or FUT confirmed with venography) by 68%, (from<br />
25% [95% CI: 24-27%] to 8% [95% CI: 7-8%]). 2 LMWH has comparable efficacy to LDUH for<br />
prevention <strong>of</strong> VTE, <strong>and</strong> may be more effective for preventing proximal DVT <strong>and</strong> PE. 2,15,16<br />
Pooled results (though heterogeneity was present) from 26 studies showed that high dose<br />
LMWH (>3400 anti-Xa units) does not reduce DVT more than low dose LMWH does (
Orthopedic <strong>Patient</strong>s<br />
All studies included in the evaluation <strong>of</strong> prophylaxis for orthopedic patients used<br />
venography to diagnose DVT.<br />
Total Hip Replacement<br />
Within 7-14 days after total hip replacement (THR), patients not receiving prophylaxis<br />
have an incidence <strong>of</strong> total <strong>and</strong> proximal DVT <strong>of</strong> 54% <strong>and</strong> 25%, respectively. Asymptomatic PE<br />
may occur in 5-15%, symptomatic PE in 1-2%, <strong>and</strong> fatal PE in 0.1-0.4%. 2,21 In patients<br />
undergoing THR, many RCTs have shown that both mechanical <strong>and</strong> pharmacologic measures<br />
are highly effective in the prevention <strong>of</strong> VTE.<br />
A meta-analysis <strong>of</strong> 52 RCTs showed that patients receiving prophylaxis with LMWH,<br />
LDUH, warfarin, aspirin or IPC had fewer total DVTs (proximal plus distal). 21 IPC <strong>and</strong> ES<br />
reduce distal DVT but do not significantly reduce proximal DVT in these patients. 2,17 Compared<br />
with placebo, prophylaxis with warfarin or LMWH resulted in the greatest reduction in proximal<br />
DVT: a reduction <strong>of</strong> 70-80% with either method (risk <strong>of</strong> proximal DVT with warfarin 5-6%, RR<br />
6.3% [95% CI: 4.7-8.4%]; with LMWH 6–8%, RR 7.7% [95% CI: 5.7-10.3%]). 2,21 Both LMWH<br />
<strong>and</strong> warfarin resulted in significantly fewer proximal DVTs compared with LDUH or IPC<br />
(p
data show that aspirin (6 studies, 443 patients), warfarin (9 studies, 1294 patients), <strong>and</strong> LDUH (2<br />
studies, 172 patients) do not significantly reduce proximal DVT following TKR. 2 <strong>Patient</strong>s given<br />
warfarin prophylaxis who were routinely screened with venography had an 8–12% risk <strong>of</strong><br />
proximal DVT. However, one study that followed patients for 3 months following TKR found<br />
that patients who received warfarin had a rate <strong>of</strong> symptomatic VTE <strong>of</strong> only 0.8%. 24<br />
The incidence <strong>of</strong> PE is low in TKR patients. Studies in which PE was diagnosed by lung<br />
scan or angiography showed the rate <strong>of</strong> symptomatic <strong>and</strong> asymptomatic PE in patients treated<br />
with aspirin to be 1.3% <strong>and</strong> 11.7%, respectively, with warfarin 0.4% <strong>and</strong> 8.2%, <strong>and</strong> with IPC<br />
0.5% <strong>and</strong> 6.3%. Symptomatic PE occurred in none <strong>of</strong> 177 patients receiving LMWH. No studies<br />
<strong>of</strong> LMWH used routine lung scanning. 23<br />
Initiation <strong>of</strong> Therapy<br />
Meta-analysis <strong>of</strong> patients undergoing THR found that LMWH initiated preoperatively<br />
resulted in a lower risk <strong>of</strong> VTE than LMWH started postoperatively (10% vs. 15%, p=0.02). 25<br />
Major bleeding occurred less frequently in the group receiving preoperative LMWH (0.9% vs.<br />
3.5%, p=0.01).<br />
Duration <strong>of</strong> Therapy<br />
The appropriate duration <strong>of</strong> VTE prophylaxis following orthopedic surgery is not clearly<br />
established. However, it is clear that the increased risk <strong>of</strong> DVT persists post-discharge. In the<br />
largest r<strong>and</strong>omized trial <strong>of</strong> post-discharge LMWH, 533 patients received either 35 days <strong>of</strong><br />
dalteparin or 6 days <strong>of</strong> warfarin followed by placebo. <strong>Patient</strong>s treated with extended LMWH had<br />
significantly fewer total <strong>and</strong> proximal DVTs from day 6 to day 35 (for all DVT, dalteparin 4.8%<br />
vs. placebo 10.5%, p=0.03; for proximal DVT, dalteparin 1% vs. placebo 4.8%, p=0.02), as well<br />
as from day 1 to day 35 (dalteparin 2.6% vs. warfarin/placebo 9.2%, p=0.002). Seventy five<br />
percent <strong>of</strong> these DVTs were asymptomatic. Symptomatic thrombi occurred in approximately 1-<br />
2%. No patient had symptomatic, objectively documented pulmonary embolism. 26 Pooled data<br />
from this study <strong>and</strong> 5 others that compared in-hospital LMWH followed by LMWH or placebo<br />
found that prolonged LMWH resulted in a 66% reduction in total DVT (14% vs. 27%) <strong>and</strong> a<br />
66% reduction in proximal DVT (4% vs. 12%) by 35 days. 2<br />
Neurosurgery<br />
<strong>Patient</strong>s undergoing neurosurgical procedures carry a risk <strong>of</strong> developing DVT <strong>of</strong><br />
approximately 22%, <strong>and</strong> a risk <strong>of</strong> proximal DVT <strong>of</strong> 5%. 2 The risk <strong>of</strong> DVT is increased in<br />
patients undergoing intracranial surgery compared to spinal surgery. Among patients undergoing<br />
intracranial surgery, those with malignant tumors have a higher risk <strong>of</strong> DVT than those with<br />
benign tumors. Increasing age <strong>and</strong> increasing duration <strong>of</strong> neurosurgery further increase the risk<br />
<strong>of</strong> VTE. Meta-analysis <strong>of</strong> r<strong>and</strong>omized studies comparing LMWH to placebo (one study) or<br />
LMWH plus ES to ES alone (2 studies) in neurosurgical patients found that LMWH was<br />
associated with a 38% reduction in total DVT (18% vs. 28%, p
Trauma<br />
Trauma patients, especially those with orthopedic injuries, are at very high risk for VTE.<br />
DVT occurs in over 50% <strong>of</strong> these patients, proximal DVT in approximately 20%, <strong>and</strong> fatal PE in<br />
up to 2%. 2 Few r<strong>and</strong>omized trials have studied VTE prophylaxis in trauma patients. Metaanalysis<br />
shows that VTE prophylaxis, either with LDUH (4 r<strong>and</strong>omized trials, OR 0.97, 95% CI:<br />
0.35-2.6 for LDUH vs. control) or ES (3 r<strong>and</strong>omized trials, OR 0.77, 95% CI: 0.27-2.2), did not<br />
reduce DVT in trauma patients compared with placebo. 29 Heterogeneity was present for all<br />
comparisons due to differences in methods used to diagnose DVT <strong>and</strong> differences in trauma<br />
populations. Pooled data from 4 studies (2 r<strong>and</strong>omized <strong>and</strong> 2 nonr<strong>and</strong>omized) <strong>of</strong> LDUH versus<br />
mechanical prophylaxis found no difference in risk <strong>of</strong> DVT (OR 1.16, 95% CI: 0.5-2.7). 29 One<br />
r<strong>and</strong>omized trial comparing LDUH with LMWH in trauma patients screened with venography<br />
found that DVT was reduced by 30% with enoxaparin (31% vs. 44%, p=0.01). 30 Of the 265<br />
patients r<strong>and</strong>omized, only one patient (in the LMWH group) suffered a PE.<br />
Acute Spinal Cord Injury<br />
The risk <strong>of</strong> DVT (diagnosed by FUT or impedance plethysmography) in patients with<br />
acute spinal cord injury ranges from 40-90%. 2 The only trial using screening venography in<br />
patients with acute spinal cord injury who were not receiving prophylaxis found DVT in 81%<br />
<strong>and</strong> proximal DVT in 35%. 31<br />
Prophylaxis in acute spinal cord injury has not been well studied. In the only study<br />
evaluating use <strong>of</strong> IPC, proximal DVT occurred in 40% <strong>of</strong> patients with IPC alone, <strong>and</strong> in 25% <strong>of</strong><br />
patients with IPC combined with aspirin <strong>and</strong> dipyridamole. 32 One r<strong>and</strong>omized trial with 35<br />
patients found that LMWH (tinzaparin 3500U daily) compared to LDUH (5000U three times a<br />
day) reduced DVT from 16% to 0%, a reduction that did not reach statistical significance. 33<br />
However, a later prospective cohort study diagnosed DVT in 13% <strong>of</strong> 48 patients prophylactically<br />
treated with LMWH. 34 In both studies, screening for DVT was done with ultrasound or<br />
ultrasound plus impedance plethysmography, with confirmation by venography. Of 15 acute<br />
spinal cord injury patients from a r<strong>and</strong>omized trial <strong>of</strong> major trauma patients, DVT was detected<br />
in 67% <strong>of</strong> those given LDUH <strong>and</strong> 50% with LMWH. Proximal DVT occurred in 13% receiving<br />
LDUH <strong>and</strong> none prophylaxed with LMWH. 30<br />
338
<strong>Medical</strong> <strong>Patient</strong>s<br />
VTE prevention in hospitalized medical patients has not been studied as extensively as in<br />
surgical patients. DVT occurs in 24% <strong>of</strong> patients with myocardial infarction, <strong>and</strong> is reduced by<br />
71% with LDUH compared with placebo with no increase in bleeding (4 studies with 165<br />
patients). 2 However, most patients with acute myocardial infarction receive full anticoagulation<br />
for treatment <strong>of</strong> the acute coronary syndrome.<br />
After ischemic stroke, 55% <strong>of</strong> untreated patients develop DVT. Prophylaxis with either<br />
LDUH or LMWH given for 10-14 days reduces DVT by 56%, from 55% (95% CI: 49%-60%) to<br />
24% (95% CI: 20%-29%). Two studies directly comparing LDUH (5000 U three times daily) to<br />
LMWH (enoxaparin 40 mg once daily), using venography for diagnosis, found greater reduction<br />
in DVT with LMWH. 2<br />
Among hospitalized patients with other medical conditions, the rate <strong>of</strong> DVT is<br />
approximately 16%. 2 A meta-analysis <strong>of</strong> studies <strong>of</strong> hospitalized patients with conditions other<br />
than myocardial infarction or ischemic stroke given VTE prophylaxis with unfractionated or low<br />
molecular weight heparin showed a 56% reduction in DVT (RR 0.44, 95% CI: 0.29-0.64) <strong>and</strong> a<br />
52% reduction in PE (RR 0.48, 95% CI: 0.34-0.68). No significant difference was found between<br />
LMWH <strong>and</strong> LDUH in incidence <strong>of</strong> DVT, PE, or mortality; however, major hemorrhage was<br />
lower with LMWH than with LDUH (RR 0.48, 95% CI: 0.23-1.00). 35<br />
Another r<strong>and</strong>omized trial comparing enoxaparin with placebo in 866 medical patients<br />
found a 63% reduction in overall VTE risk with enoxaparin (40 mg) compared with placebo in<br />
the first 14 days (5.5% vs. 15%, RR 0.37, 95% CI 0.22-0.63). Proximal DVT was reduced by<br />
65% (1.7% vs. 4.9%, RR 0.40, 95% CI 0.23-0.69). Significant reductions in total <strong>and</strong> proximal<br />
DVT persisted at 110 days. VTE was not reduced in the group receiving 20 mg <strong>of</strong> enoxaparin.<br />
The most common medical conditions were acute infectious disease, acute respiratory failure,<br />
New York Heart Association Class III or IV congestive heart failure. 36<br />
The risk <strong>of</strong> VTE is higher in patients with malignant disease, especially those with<br />
adenocarcinoma or brain tumors. Factors associated with increased VTE in cancer patients<br />
include chemotherapy, surgery, <strong>and</strong> indwelling central venous catheters. Breast cancer patients<br />
treated with tamoxifen also have higher rates <strong>of</strong> VTE. In patients with indwelling central venous<br />
catheters who received low dose warfarin (1 mg per day), upper extremity DVT was reduced by<br />
75% (9.5% vs. 37.5%). 37<br />
In summary, for general surgery patients who are at moderate risk or greater for VTE,<br />
LDUH <strong>and</strong> LMWH are the most effective methods for prophylaxis. IPC also provides effective<br />
DVT prevention, but in very high risk patients should only be used in conjunction with heparin.<br />
For orthopedic procedures, LMWH <strong>and</strong> warfarin are the most effective preventive measures.<br />
Neurosurgical patients should receive LMWH or LDUH, while in acute spinal cord injury or<br />
trauma, LMWH provides the largest reduction in VTE. For stroke or medical patients, LMWH<br />
<strong>and</strong> LDUH show the greatest benefit for VTE prevention (see Table 31.2).<br />
Summary <strong>of</strong> Prophylaxis Recommendations<br />
For general surgery patients at moderate to high risk, LDUH <strong>and</strong> LMWH have<br />
comparable effectiveness for prevention <strong>of</strong> VTE, with similar bleeding risks. LDUH is generally<br />
more cost-effective. For high risk patients, IPC has not been consistently shown to prevent<br />
proximal DVT (See Table 31.3).<br />
For major orthopedic procedures (THR, TKR, hip fracture repair), prophylaxis with<br />
LMWH or warfarin results in the greatest benefit, with a small increase in bleeds compared with<br />
339
no prophylaxis, but no difference between the two agents. Warfarin may be more cost-effective,<br />
but necessitates monitoring <strong>and</strong> dose adjustment. This should be considered in choosing between<br />
the two agents.<br />
IPC, LMWH, <strong>and</strong> LDUH are all acceptable methods <strong>of</strong> prophylaxis for neurosurgical<br />
patients. Each effectively reduces proximal DVT with no increase in major bleeding. There are<br />
no data on cost-effectiveness <strong>of</strong> prophylaxis for these patients.<br />
Data for prophylaxis <strong>of</strong> trauma patients does not show conclusive VTE reduction with<br />
any agent. However, the risk <strong>of</strong> VTE among these patients is high <strong>and</strong> prophylaxis should be<br />
considered, especially for those with orthopedic injuries. LMWH is a safe method <strong>of</strong> prophylaxis<br />
<strong>and</strong> has not been shown to increase major bleeds in blunt trauma patients. However if a patient is<br />
at high risk for major bleeding, IPC should be considered.<br />
For medical patients, LDUH <strong>and</strong> LMWH are both effective for reducing VTE. LDUH<br />
may result in slightly more bleeding, but is more cost-effective. <strong>Patient</strong>s at high risk for bleeding<br />
should be given prophylaxis with ES or IPC.<br />
Potential for Harm<br />
There is no documented risk associated with mechanical devices, although there is a<br />
potential risk that patients’ legs will be examined less frequently. The major risk <strong>of</strong><br />
pharmacologic prophylaxis is bleeding. Bleeding is typically considered major when the<br />
hemoglobin decreases by at least 2 g/dL, when red blood cell transfusion is necessary, when<br />
intracranial or retroperitoneal bleeding occurs, or when bleeding requires surgical intervention.<br />
Another consequence <strong>of</strong> heparin therapy may be thrombocytopenia.<br />
For general surgery, there is no difference in the risk <strong>of</strong> major bleeding between LMWH<br />
<strong>and</strong> LDUH, although fewer minor bleeds may occur with LMWH (RR 0.64, 95% CI 0.58-<br />
0.70). 15 However, when evaluated according to dose <strong>of</strong> LMWH, compared with LDUH, lowdose<br />
LMWH (
LMWH was more cost-effective than LDUH. 39 An economic evaluation using decision analysis<br />
compared LDUH <strong>and</strong> LMWH for patients undergoing colorectal surgery. There was no<br />
difference in risk <strong>of</strong> VTE between groups. Per 1000 patients treated, prophylaxis with<br />
enoxaparin compared with LDUH resulted in 12 excess major bleeds <strong>and</strong> an additional cost <strong>of</strong><br />
$145,667. 40 This supports LDUH as a more cost-effective measure for patients undergoing<br />
general surgery.<br />
Meta-analysis <strong>of</strong> studies <strong>of</strong> the cost-effectiveness <strong>of</strong> VTE prophylaxis for patients<br />
undergoing hip arthroplasty found that with a 2.6 to 1 price ratio between LMWH <strong>and</strong> LDUH,<br />
use <strong>of</strong> LMWH would save the health care system approximately $50,000 per 1000 patients<br />
treated. 41 For VTE prophylaxis after TKR, LMWH was more cost-effective than warfarin, saving<br />
$2401 per 100 patients ($9197 vs. $11,598 per 100 patients). 42 This study was done in Canada,<br />
where LMWH is less costly than in the United States. Another meta-analysis <strong>of</strong> THR patients<br />
found that LDUH would decrease the cost <strong>of</strong> care related to DVT by $200 per patient. Compared<br />
with warfarin, LMWH would be more effective in preventing DVT (expected DVT rate<br />
420/10,000 with warfarin <strong>and</strong> 250/10,000 with LMWH), <strong>and</strong> death from VTE (110/10,000 with<br />
warfarin <strong>and</strong> 70/10,000 with LMWH). However, preventing one death with LMWH use instead<br />
<strong>of</strong> warfarin would cost approximately $12,000. 43<br />
In medical patients, LDUH given 3 times daily was found to be more cost-effective than<br />
LMWH, with a savings per 1000 patients <strong>of</strong> $10,753 compared with enoxaparin 40 mg/d, <strong>and</strong><br />
$15,000 compared with enoxaparin 20 mg/d. 44 The higher cost associated with enoxaparin 20<br />
mg/d results from the higher incidence <strong>of</strong> complications with this regimen. This supports use <strong>of</strong><br />
LDUH as the preferred method <strong>of</strong> prophylaxis for medical patients at the present time, though<br />
this is an area <strong>of</strong> active investigation. No studies were found evaluating the cost <strong>of</strong> DVT<br />
prophylaxis in neurosurgery patients.<br />
Comment<br />
As noted earlier, despite the clear evidence <strong>of</strong> effectiveness, DVT prophylaxis is<br />
underused. The reasons for this underuse have not been completely elucidated, nor have the<br />
optimal strategies for improving prophylaxis been fully identified. Various strategies have been<br />
tried in an effort to improve physician utilization <strong>of</strong> appropriate VTE prophylaxis. One hospital<br />
studied the impact <strong>of</strong> educational programs promoting guidelines for prophylaxis. It found that<br />
presentations to staff, distributions <strong>of</strong> cards with the hospital’s guidelines, <strong>and</strong> posters increased<br />
prophylaxis <strong>of</strong> surgical patients from 59% to 70%, <strong>and</strong> to 77% for high-risk patients (see also<br />
Chapters 51 <strong>and</strong> 54). 45<br />
In another study, a computer-based clinical decision support system (CDSS) (see Chapter<br />
53) providing information pertaining to VTE prophylaxis was used in an orthopedic surgery<br />
department <strong>of</strong> a teaching hospital. The investigators monitored the impact <strong>of</strong> CDSS use in 1971<br />
patients undergoing orthopedic surgery. Compliance with guidelines was 83% during control<br />
periods <strong>and</strong> 95% during intervention periods. Inappropriate practice decisions occurred almost 4<br />
times more frequently during control versus intervention periods. 46<br />
A third study evaluating methods to improve VTE prophylaxis among intensive care unit<br />
patients found that staff education improved use <strong>of</strong> appropriate prophylaxis from 38% to 62%.<br />
When required computerized order sets were added to education, appropriate prophylaxis<br />
increased to 97%. 47<br />
These studies suggest that either a knowledge gap or lack <strong>of</strong> awareness may exist among<br />
practitioners. For institutions or groups attempting to improve appropriate use <strong>of</strong> measures to<br />
prevent VTE, guidelines made available via computerized support systems or order sets provide<br />
341
the most effective means <strong>of</strong> implementing appropriate VTE prophylaxis, especially when these<br />
systems are linked to effective educational programs.<br />
342
Table 31.1. Mechanical <strong>and</strong> pharmacologic preventative measures for VTE<br />
Practice Type Description Comment<br />
Graduated Elastic<br />
Stockings (ES)<br />
Intermittent<br />
pneumatic<br />
compression (IPC)<br />
Mechanical Fitted hose that extend<br />
above the knee<br />
Mechanical Devices fitted over lower<br />
extremities that<br />
sequentially inflate <strong>and</strong><br />
deflate<br />
Aspirin Pharmacologic Usually 325 mg/d<br />
Warfarin Pharmacologic 5-10 mg started the day<br />
<strong>of</strong> or after surgery; adjust<br />
to achieve an INR <strong>of</strong> 2-3<br />
Low-dose<br />
unfractionated<br />
heparin (LDUH)<br />
Low Molecular<br />
Weight Heparin<br />
(LMWH)<br />
Pharmacologic Generally 5000 U<br />
subcutaneous bid or tid,<br />
though some studies have<br />
adjusted dose to maintain<br />
PTT at high end <strong>of</strong><br />
normal<br />
Pharmacologic Dose depends on type <strong>of</strong><br />
surgery <strong>and</strong> VTE risk*<br />
343<br />
Fitted hose are more<br />
efficacious than non-fitted<br />
Monitoring <strong>of</strong> INR needed<br />
Contraindicated if active<br />
bleeding or history <strong>of</strong><br />
thrombocytopenia; no<br />
need to follow coagulation<br />
studies (unless adjusted<br />
dose is used)<br />
No need to monitor<br />
coagulation studies<br />
* LMWH dosing: Enoxaparin 20 mg SC daily (moderate risk surgery) or 40 mg SC daily (can go<br />
up to 30 mg SC q12h for high risk general surgery, major trauma or acute spinal cord injury);<br />
dalteparin 2500–5000 U SC daily; nadroparin 2500 U SC daily; tinzaparin 3500-4500 U SC<br />
daily (may be dosed 75U/kg/d for orthopedic surgery).
Table 31.2. Summary <strong>of</strong> DVT risk <strong>and</strong> prophylactic methods providing significant risk<br />
reduction*<br />
Surgery/<br />
Condition<br />
General<br />
Surgery 2<br />
THR 2<br />
TKR 2,23<br />
Neurosurgery<br />
27,28<br />
Trauma 2,30<br />
Risk <strong>of</strong> all DVT in<br />
untreated patients<br />
25%<br />
54%<br />
64%<br />
28%<br />
Type <strong>of</strong><br />
Prophylaxis<br />
30-60% LMWH<br />
Acute Spinal<br />
Cord Injury 2 80% Not established<br />
Ischemic<br />
stroke 2<br />
55%<br />
344<br />
Risk Reduction with<br />
Prophylaxis<br />
ES 44% 3<br />
LDUH 68% 47<br />
LMWH 76% 21<br />
IPC 88% 2<br />
LMWH 70% 30<br />
warfarin 59% 13<br />
LMWH 52% 13<br />
IPC 73% 6<br />
LMWH 38% 3<br />
LDUH 72% 1†<br />
30%<br />
(compared to LDUH)<br />
LDUH 56% 5<br />
LMWH 58% 3<br />
Danaparoid 82% 4<br />
Number <strong>of</strong><br />
Studies<br />
76% 2†<br />
<strong>Medical</strong><br />
LMWH<br />
39% 2<br />
conditions 2 16%<br />
LDUH 61% 3†<br />
* DVT indicates deep venous thrombosis; ES, graduated elastic stockings; IPC, intermittent<br />
pneumatic compression; LDUH, low-dose unfractionated heparin; LMWH, low molecular<br />
weight heparin; THR, total hip replacement; <strong>and</strong> TKR, total knee replacement.<br />
† DVT diagnosed by fibrinogen uptake test (FUT)<br />
1
Table 31.3. Recommended VTE prophylaxis for surgical procedures <strong>and</strong> medical<br />
conditions*<br />
Surgery/Condition Recommended<br />
Prophylaxis<br />
General Surgery – low-risk: minor<br />
procedures,
References<br />
1. US Bureau <strong>of</strong> the Census. Statistical Abstract <strong>of</strong> the United States: 2000. 120th edition.<br />
Washington, DC. 2000.<br />
2. Geerts WH, Heit JA, Clagett GP, Pineo GF, Colwell CW, Anderson FA, et al. Prevention<br />
<strong>of</strong> venous thromboembolism. Chest. 2001;119:132S-175S.<br />
3. Caprini JA, Arcelus JI, H<strong>of</strong>fman K, Mattern T, Laubach M, Size GP, et al. Prevention <strong>of</strong><br />
venous thromboembolism in North America: results <strong>of</strong> a survey among general surgeons. J<br />
Vasc Surg. 1994;20:751-758.<br />
4. Janku GV, Paiement GD, Green HD. Prevention <strong>of</strong> venous thromboembolism in<br />
orthopaedics in the United States. Clin Ortho & Related Research. 1996:313-321.<br />
5. Bratzler DW, Raskob GE, Murray CK, Bumpus LJ, Piatt DS. Underuse <strong>of</strong> venous<br />
thromboembolism prophylaxis for general surgery patients: physician practices in the<br />
community hospital setting. Arch Intern Med. 1998;158:1909-1912.<br />
6. Stratton MA, Anderson FA, Bussey HI, Caprini J. Prevention <strong>of</strong> venous thromboembolism:<br />
adherence to the 1995 American College <strong>of</strong> Chest Physicians Consensus Guidelines for<br />
Surgical <strong>Patient</strong>s. Arch Intern Med. 2000;160:334-340.<br />
7. Clagett GP, Anderson FAJ, Levine MN, Wheeler HB. Prevention <strong>of</strong> venous<br />
thromboembolism. Chest. 1995;108S:312S-334S.<br />
8. Keane MG, P IE, Goldhaber SZ. Utilization <strong>of</strong> venous thromboembolism prophylaxis in<br />
the medial intensive care unit. Chest. 1994;106:13-14.<br />
9. Goldhaber SZ, Dunn K, MacDougall RC. New onset <strong>of</strong> venous thromboembolism among<br />
hospitalized patients at Brigham <strong>and</strong> Women's Hospital is caused more <strong>of</strong>ten by<br />
prophylaxis failure than by withholding treatment. Chest. 2000;118:1680-1684.<br />
10. Heijboer H, Cogo A, Buller HR. Detection <strong>of</strong> deep vein thrombosis with impedance<br />
plethysmography <strong>and</strong> real-time compression ultrasonography in hospitalized patients. Arch<br />
Intern Med. 1992;152:1901-1903.<br />
11. Lensing AW, Hirsh J. 1251-fibrinogen leg scanning: reassessment <strong>of</strong> its role for the<br />
diagnosis <strong>of</strong> venous thrombosis in post-operative patients. Thromb Heamost. 1993;69:2-7.<br />
12. Esk<strong>and</strong>ari MK, Sugimoto H, Richardson T. Is color-flow duplex a good diagnostic test for<br />
detection <strong>of</strong> isolated calf vein thrombosis in high-risk patients? Angiology. 2000;51:705-<br />
710.<br />
13. Perrier A, Miron MJ, Desmarais S, de Moerloose P. Using clinical evaluation <strong>and</strong> lung<br />
scan to rule out suspected pulmonary embolism. Arch Intern Med. 2000;160:512-516.<br />
14. Meignan M, Rosso J, Gauthier H, Brunengo F. Systematic lung scans reveal a high<br />
frequency <strong>of</strong> silent pulmonary embolism in patients with proximal deep venous<br />
thrombosis. Arch Intern Med. 2000;160:159-164.<br />
15. Palmer AJ, Schramm W, Kirchh<strong>of</strong> B, Bergemann R. Low molecular weight heparin <strong>and</strong><br />
unfractionated heparin for prevention <strong>of</strong> thrombo-embolism in general surgery: a metaanalysis<br />
<strong>of</strong> r<strong>and</strong>omised clinical trials. Haemostasis. 1997;27:65-74.<br />
16. Koch A, Bouges S, Ziegler S, Dinkel H, Daures JP, Victor N. Low molecular weight<br />
heparin <strong>and</strong> unfractionated heparin in thrombosis prophylaxis after major surgical<br />
intervention: update <strong>of</strong> previous meta-analyses. Br J Surg. 1997;84:750-759.<br />
346
17. Vanek VW. Meta-analysis <strong>of</strong> effectiveness <strong>of</strong> intermittent pneumatic compression devices<br />
with a comparison <strong>of</strong> thigh-high to knee-high sleeves. American Surgeon. 1998;64:1050-<br />
1058.<br />
18. Wells PS, Lensing AW, Hirsh J. Graduated compression stockings in the prevention <strong>of</strong><br />
postoperative venous thromboembolism. A meta-analysis. Arch Intern Med. 1994;154:67-<br />
72.<br />
19. Agu O, Hamilton G, Baker D. Graduated compression stockings in the prevention <strong>of</strong><br />
venous thromboembolism. Br J Surg. 1999;86:992-1004.<br />
20. Amarigiri SV, Lees TA. Elastic compression stockings for prevention <strong>of</strong> deep vein<br />
thrombosis. In: The Cochrane Library, Issue1, 2001.<br />
21. Freedman KB, Brookenthal KR, Fitzgerald RH, Jr., Williams S, Lonner JH. A metaanalysis<br />
<strong>of</strong> thromboembolic prophylaxis following elective total hip arthroplasty. J Bone<br />
Joint Surg. 2000;82-A:929-938.<br />
22. Pulmonary Embolism Prevention (PEP) Trial Collaborative Group. Prevention <strong>of</strong><br />
pulmonary embolism <strong>and</strong> deep vein thrombosis with low dose aspirin: Pulmonary<br />
Embolism Prevention (PEP) trial. Lancet. 2000;355:1295-1302.<br />
23. Westrich GH, Haas SB, Mosca P, Peterson M. Meta-analysis <strong>of</strong> thromboembolic<br />
prophylaxis after total knee arthroplasty. J Bone Joint Surg. 2000;82-B:795-800.<br />
24. Heit JA, Berkowitz SD, Bona R, Cabanas V, Corson JD, Elliott CG, et al. Efficacy <strong>and</strong><br />
safety <strong>of</strong> low molecular weight heparin compared to warfarin for the prevention <strong>of</strong> VTE<br />
after total knee replacement surgery: a double-blind, dose-ranging study. Thromb Haemost.<br />
1997;77:32-38.<br />
25. Hull RD, Brant RF, Pineo GF, Stein PD. Preoperative vs postoperative initiatin <strong>of</strong> lowmolecular-weight<br />
heparin prophylaxis against venous thromboembolism in patients<br />
undergoing elective hip replacement. Arch Intern Med. 1999;159:137-141.<br />
26. Hull RD, Pineo GF, Francis C, Bergqvist D, Fellenius C, Soderberg K, et al. Lowmolecular-weight<br />
heparin prophylaxis using dalteparin extended out-<strong>of</strong>-hospital placebo in<br />
hip arthroplasty patients. Arch Intern Med. 2000;160:2208-2215.<br />
27. Iorio A, Agnelli G. Low-molecular-weight <strong>and</strong> unfractionated heparin for prevention <strong>of</strong><br />
venous thromboembolism in neurosurgery: a meta-analysis. Arch Intern Med.<br />
2000;160:2327-2332.<br />
28. Cerrato D, Ariano C, Fiacchino F. Deep vein thrombosis <strong>and</strong> low-dose heparin prophylaxis<br />
in neurosurgical patients. J Neurosurg. 1978;49:378-381.<br />
29. Velmahos GC, Kern J, Chan LS, Oder D, Murray JA, Shekelle P. Prevention <strong>of</strong> venous<br />
thromboembolism after injury: an evidence-based report–part I: analysis <strong>of</strong> risk factors <strong>and</strong><br />
evaluation <strong>of</strong> the role <strong>of</strong> vena caval filters. J Trauma. 2000;49:132-138.<br />
30. Geerts WH, Jay RM, Code KI. A comparison <strong>of</strong> low dose heparin with low molecular<br />
weight heparin as prophylaxis against venous thromboembolism after major trauma. N<br />
Engl J Med. 1996;335:701-707.<br />
31. Geerts WH, Code KI, Jay RM, Chen E. A prospective study <strong>of</strong> venous thromboembolism<br />
after major trauma. N Engl J Med. 1994;331:1601-1606.<br />
32. Green D, Rossi EC, Yao JS. Deep vein thrombosis in spinal cord injury: effect <strong>of</strong><br />
prophylaxis with calf compression, aspirin <strong>and</strong> dipyridamole. Paraplegia. 1982;20:227-<br />
234.<br />
33. Green D, Lee MY, Lim AC. Prevention <strong>of</strong> thromboembolism after spinal cord injury using<br />
low molecular weight heparin. Ann Intern Med. 1990;113:571-574.<br />
347
34. Green D, Chen D, Chmiel JS. Prevention <strong>of</strong> thromboembolism in spinal cord injury: role <strong>of</strong><br />
low molecular weight heparin. Arch Phys Med Rehabil. 1994;75:290-292.<br />
35. Mismetti P, Laporte-Simitsidis S, Tardy B, Cucherat M, Buchmuller A, Juillard-Delsart D,<br />
et al. Prevention <strong>of</strong> venous thromboembolism in internal medicine with unfractionated or<br />
low-molecular-weight heparins: a meta-analysis <strong>of</strong> r<strong>and</strong>omised clinical trials. Thromb<br />
Haemost. 2000;83:14-19.<br />
36. Samama MM, Cohen AT, Darmon JY, Desjardins L, Eldor A, Janbon C, et al. A<br />
comparison <strong>of</strong> enoxaparin with placebo for the prevention <strong>of</strong> VTE in acutely ill medical<br />
patients. N Engl J Med. 1999;341:793-800.<br />
37. Bern MM, Lokich JJ, Wallach SR. Very low doses <strong>of</strong> warfarin can prevent thrombosis in<br />
central venous catheters: a r<strong>and</strong>omized prospective trial. Ann Intern Med. 1990;112:423-<br />
428.<br />
38. Norwwod SH, McAuley CE, Berne JD, Vallina VL, Kerns DB, Grahm TW, et al. A<br />
potentially exp<strong>and</strong>ed role for enoxaparin in preventing venous thromboembolism in high<br />
risk blunt trauma patients. J Am Coll Surg. 2001;192:161-167.<br />
39. Bergqvist D, Lindgren B, Matzsch T. Comparison <strong>of</strong> the cost <strong>of</strong> preventing postoperative<br />
deep vein thrombosis with either unfractionated or low molecular weight heparin. Br J<br />
Surg. 1996;83:1548-1552.<br />
40. Etchells E, McLeod RS, Geerts W, Barton P, Detsky AS. Economic analysis <strong>of</strong> low-dose<br />
heparin vs the low-molecular-weight heparin enoxaparin for prevention <strong>of</strong> venous<br />
thromboembolism after colorectal surgery. Arch Intern Med. 1999;159:1221-1228.<br />
41. Anderson DR, O'Brien BJ, Levine MN, Roberts R, Wells PS, Hirsh J. Efficacy <strong>and</strong> cost <strong>of</strong><br />
low-molecular-weight heparin compared with st<strong>and</strong>ard heparin for the prevention <strong>of</strong> deep<br />
vein thrombosis after total hip arthroplasty. Ann Intern Med. 1993;119:1105-1112.<br />
42. Hull RD, Raskob GE, Pineo GF, Feldstein W. Subcutaneous low-molecular-weight heparin<br />
vs warfarin for prophylaxis <strong>of</strong> deep vein thrombosis after hip or knee implantation: an<br />
economic perspective. Arch Intern Med. 1997;157:293-303.<br />
43. Menzin J, Colditz GA, Regan MM, Richner RE, Oster G. Cost-effectiveness <strong>of</strong> enoxaparin<br />
vs low-dose warfarin in the prevention <strong>of</strong> deep-vein thrombosis after total hip replacement<br />
surgery. Arch Intern Med. 1995;155:757-764.<br />
44. Wade WE. Cost-effectiveness analysis <strong>of</strong> deep vein thrombosis prophylaxis in internal<br />
medicine patients. Thrombrosis Research. 1999;94:65-68.<br />
45. Peterson GM, Drake CI, Jupe DM, Vial JH, Wilkinson S. Educational campaign to<br />
improve the prevention <strong>of</strong> postoperative venous thromboembolism. J Clin Pharm<br />
Therapeutics. 1999;24:279-287.<br />
46. Durieux P, Nizard R, Ravaud P, Mounier N, Lepage E. A clinical decision support system<br />
for prevention <strong>of</strong> venous thromboembolism. JAMA. 2000;283:2816-2821.<br />
47. Levi D, Kupfter Y, Seneviratne C, Tessler S. Computerized order entry sets <strong>and</strong> intensive<br />
education improve the rate <strong>of</strong> prophylaxis for deep vein thrombophlebitis. Chest.<br />
1998;114S:280S.<br />
348
Chapter 32. Prevention <strong>of</strong> Contrast-Induced Nephropathy<br />
Lorenzo Di Francesco, MD<br />
Mark V. Williams, MD<br />
Emory University School <strong>of</strong> Medicine<br />
Background<br />
Radiocontrast-induced nephropathy (RCIN) represents an increasingly common cause <strong>of</strong><br />
treatment-related renal failure 1-3 <strong>and</strong> increases mortality independent <strong>of</strong> other risk factors. 4 Major<br />
risk factors for RCIN include chronic renal insufficiency, 2,3,5 diabetes mellitus 2,3,5 (especially<br />
when accompanied by renal insufficiency 1 ), any condition associated with decreased effective<br />
circulating volume, 6 <strong>and</strong> use <strong>of</strong> large doses <strong>of</strong> contrast media. 2,3,5,6<br />
For at-risk patients, clinicians must use their judgment to determine if imaging modalities<br />
that do not involve contrast media are an acceptable alternative to contrast studies. In many<br />
cases, however, such alternatives do not exist. Moreover, RCIN occurs in patients without<br />
obvious risk factors. Thus, strategies for reducing the incidence <strong>of</strong> RCIN include not just risk<br />
factor identification, but modification <strong>of</strong> these risk factors, choice <strong>of</strong> contrast media less likely to<br />
cause RCIN, <strong>and</strong> administration <strong>of</strong> therapeutic agents that further reduce the risk <strong>of</strong> RCIN.<br />
Practice Description<br />
The specific practices reviewed in this chapter are:<br />
• Use <strong>of</strong> high versus low osmolar iodinated contrast media to prevent RCIN. 7<br />
• Use <strong>of</strong> a st<strong>and</strong>ard intravenous or oral hydration protocol for patients with risk<br />
factors for RCIN. 8-10 Typical intravenous protocols evaluated consist <strong>of</strong> normal saline<br />
administered at 75 mL/hr beginning at 12 hours before <strong>and</strong> ending 12 hours after the<br />
procedure. Oral protocols require ingestion <strong>of</strong> 1000 mL <strong>of</strong> water during the 10 hours<br />
prior to the procedure, followed by intravenous normal saline at 300 mL/h for 30-60<br />
minutes <strong>and</strong> continued for a total <strong>of</strong> 6 hours after the procedure.<br />
• Use <strong>of</strong> a st<strong>and</strong>ard hydration protocol supplemented by pretreatment with<br />
theophylline 11-15 (various doses <strong>and</strong> schedules)<br />
• Use <strong>of</strong> a st<strong>and</strong>ard hydration protocol supplemented by pretreatment with Nacetylcysteine<br />
16 (600 mg bid one day before <strong>and</strong> day <strong>of</strong> procedure)<br />
Single studies evaluating atrial natriuretic peptide, prostagl<strong>and</strong>in E1 17 <strong>and</strong> captopril 18<br />
were not reviewed, as the data are too preliminary, despite findings that suggest a reduction in<br />
the risk <strong>of</strong> RCIN. Although the evidence supporting the use <strong>of</strong> N-acetylcysteine largely comes<br />
from a single study as well, we do review this practice because the study was large, published in<br />
a prominent journal, <strong>and</strong> has received considerable attention among clinicians. 16<br />
The use <strong>of</strong> calcium channel blockers in preventing RCIN was not evaluated, as the<br />
existing literature predominantly indicates the practice is ineffective. 19-23<br />
349
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
While definitions <strong>of</strong> RCIN vary, most study definitions include a 25% increase in serum<br />
creatinine (SCr) <strong>and</strong>/or at least a 0.5 mg/dL increase in SCr within 48 hours <strong>of</strong> contrast<br />
administration. Using this definition, one large community-based study <strong>of</strong> 1826 patients<br />
undergoing invasive cardiac procedures reported a rate <strong>of</strong> RCIN <strong>of</strong> 14.5%. 2 A controlled<br />
prospective study <strong>of</strong> the onset <strong>of</strong> RCIN after contrast-enhanced brain CT found an incidence <strong>of</strong><br />
2.1% in low-risk patients without diabetes mellitus or chronic renal insufficiency versus 1.3% in<br />
a similar control group that did not receive any contrast (p=NS). 24 In comparison, patients in a<br />
prospective controlled study undertaken to determine the risk <strong>of</strong> nephrotoxicity from contrast<br />
radiography in patients with diabetes <strong>and</strong> renal insufficiency (SCr >1.7mg/dL) found a 9%<br />
incidence <strong>of</strong> RCIN. 1<br />
The cumulative effect <strong>of</strong> multiple risk factors increasing the risk <strong>of</strong> RCIN was<br />
demonstrated in one uncontrolled study that evaluated the effect <strong>of</strong> 5 factors (contrast volume<br />
>200 mL, albumin
the studies used aminophylline, rather than theophylline). Table 32.1 summarizes the salient<br />
features <strong>of</strong> these studies.<br />
One r<strong>and</strong>omized trial compared inpatient versus outpatient hydration regimens, 26 but we<br />
found no r<strong>and</strong>omized controlled trial that evaluated pre-hydration versus no hydration. Thus,<br />
support for the st<strong>and</strong>ard use <strong>of</strong> pre-hydration to prevent RCIN is extrapolated from r<strong>and</strong>omized<br />
controlled studies <strong>of</strong> saline versus saline plus additional pre-treatment agents like mannitol,<br />
furosemide <strong>and</strong> dopamine 8-10 <strong>and</strong> smaller observational studies 6,27,28 evaluating the benefits <strong>of</strong><br />
pre-hydration.<br />
Study Outcomes<br />
Studies evaluated Level 2 outcomes, primarily by measuring changes in serum creatinine,<br />
creatinine clearance or glomerular filtration, <strong>and</strong> assessing the frequency <strong>of</strong> developing acute<br />
renal failure after radiocontrast infusions. Most studies defined RCIN as a 25% increase in<br />
creatinine <strong>and</strong>/or at least a 0.5 mg/dL increase in serum creatinine within 48 hours <strong>of</strong> contrast<br />
administration.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
All <strong>of</strong> these studies (Table 32.1) evaluated the effects <strong>of</strong> various prophylactic measures to<br />
reduce the incidence <strong>of</strong> RCIN. Use <strong>of</strong> low-osmolar contrast media was supported by one large<br />
meta-analysis 7 that compared low versus high osmolar contrast media. Low osmolar contrast<br />
media was found to be less nephrotoxic than high osmolar contrast media, with an odds ratio for<br />
RCIN <strong>of</strong> 0.61. Among patients with baseline renal insufficiency (SCr >1.4 mg/dL) the odds ratio<br />
<strong>of</strong> developing RCIN was 0.5 if low osmolar instead <strong>of</strong> high osmolar contrast media was used.<br />
As previously noted, no r<strong>and</strong>omized controlled trials have evaluated the efficacy <strong>of</strong> prehydration<br />
versus no pre-hydration. Data from 3 r<strong>and</strong>omized controlled trials 8-10 using prehydration<br />
versus other pre-treatments <strong>and</strong> pre-hydration revealed that pre-hydration alone was<br />
equivalent to pre-hydration <strong>and</strong> low dose dopamine or mannitol, 8 <strong>and</strong>, in one study, superior to<br />
pre-hydration <strong>and</strong> furosemide. 10 The incidence <strong>of</strong> RCIN in patients with SCr >1.6 mg/dL or<br />
creatinine clearance 3 mg/dL, the incidence was only 4%. 8<br />
One retrospective, observational study <strong>of</strong> high-risk patients undergoing cardiac catheterization<br />
supports the benefit <strong>of</strong> pre-hydration (>500 mL <strong>of</strong> 0.9% NS in the pre-catheterization period,<br />
p 0.01) in reducing RCIN. 6 In addition, 2 observational studies without controls 27,28 showed<br />
that pre-hydration in high-risk patients was associated with low rates <strong>of</strong> RCIN, although one <strong>of</strong><br />
these studies 27 used a stricter definition for RCIN (increase in BUN by 50% or 20 mg/dL, <strong>and</strong>/or<br />
increase in SCr <strong>of</strong> 1 mg/dL within 24 hours).<br />
A recent study <strong>of</strong> the oral antioxidant acetylcysteine in combination with pre-hydration in<br />
high-risk patients with renal insufficiency showed significant protective effect against RCIN<br />
versus pre-hydration plus placebo. 16 This protective effect appeared to be even more significant<br />
among patients with more advanced renal dysfunction <strong>and</strong> SCr >2.5 mg/dL. The overall relative<br />
risk reduction <strong>of</strong> 90% observed in this study is so large that it raises the possibility <strong>of</strong> some sort<br />
<strong>of</strong> bias or other explanation for the observed results. Additional studies <strong>of</strong> this practice would be<br />
valuable, despite the safety <strong>and</strong> low cost <strong>of</strong> N-acetylcysteine.<br />
Studies employing theophylline are more controversial. Three r<strong>and</strong>omized control trials<br />
showed a significant protective effect <strong>of</strong> various dosages <strong>and</strong> administration routes <strong>of</strong><br />
theophylline among low-risk patients with relatively normal baseline renal function. 12-14 All 3<br />
studies showed theophylline to be protective against a decrease in glomerular filtration rate<br />
351
(GFR) or creatinine clearance (CrCl) after contrast administration. On the other h<strong>and</strong>, 2 studies<br />
conducted in high-risk patients with renal dysfunction showed no effect for theophylline in<br />
reducing RCIN. 11, 15 Thus, insufficient evidence supports the use <strong>of</strong> theophylline as prophylaxis<br />
against RCIN in high-risk patients.<br />
Potential for Harm<br />
The impact <strong>of</strong> a system to identify high-risk patients prior to contrast radiography <strong>and</strong><br />
implement aggressive prophylactic measures to reduce the incidence <strong>of</strong> RCIN has not been<br />
studied. While most patients will not experience any harm from contrast, the potential for<br />
“harm” due to delayed or cancelled investigations may be greater than the harm prevented by<br />
screening for risk factors, aggressive hydration, or use <strong>of</strong> particular pre-treatment regimens.<br />
Costs <strong>and</strong> Implementation<br />
At least 4 studies have evaluated the cost-effectiveness <strong>of</strong> low-osmolality versus highosmolality<br />
contrast media. 29-32 In all 4 studies, the selective use <strong>of</strong> low-osmolar contrast media<br />
was more cost-effective than its universal use because <strong>of</strong> the overall small benefits were<br />
outweighed by the considerable increased institutional costs. Alternatively, a st<strong>and</strong>ardized<br />
system to identify high-risk patients <strong>and</strong> implement the simple prophylactic treatment <strong>of</strong> prehydration<br />
would diminish the frequency <strong>of</strong> the target problem. It would require collaboration<br />
between the patients’ own physician <strong>and</strong> the personnel performing the particular contrast study<br />
(radiology department, radiologist, diagnostic/interventional cardiologist). This type <strong>of</strong><br />
intervention could be implemented as part <strong>of</strong> a hospital-based pathway (see Chapter 52) targeted<br />
at reducing radiocontrast-induced nephropathy.<br />
There are no cost-effectiveness or feasibility studies that evaluate protocols for<br />
aggressive identification <strong>of</strong> high-risk patients undergoing contrast radiography <strong>and</strong> utilization <strong>of</strong><br />
st<strong>and</strong>ardized hydration protocols to reduce RCIN. Two studies suggest most patients with<br />
normal renal function (SCr
Table 32.1. Studies <strong>of</strong> strategies for preventing radiocontrast-induced nephropathy<br />
(RCIN)*<br />
Study Setting Study Design,<br />
Outcomes<br />
Low osmolar contrast media<br />
Meta-analysis <strong>of</strong> the relative<br />
nephrotoxicity <strong>of</strong> high (HOCM)<br />
vs. low (LOCM) osmolar<br />
iodinated contrast media 7<br />
Pre-hydration plus diuresis<br />
<strong>Patient</strong>s with SCr >1.8mg/dL<br />
r<strong>and</strong>omized to IVF, IVF +<br />
furosemide, IVF + furosemide +<br />
low dose IV dopamine +/mannitol<br />
(if post-cardiac<br />
catheterization, PCWP 1.6mg/dL or<br />
CrCl 1.7 or CrCl<br />
1.9 mg/dL<br />
who underwent cardiac cath 6<br />
Level 1A,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
353<br />
Results<br />
LOCM less nephrotoxic than HOCM; pooled<br />
p=0.02<br />
Odds <strong>of</strong> ARF with LOCM 0.61 times that <strong>of</strong><br />
HOCM (95% CI: 0.48-0.77). <strong>Patient</strong>s with RF at<br />
baseline, odds <strong>of</strong> ARF were 0.5 (CI: 0.36-0.68).<br />
No differences in rates <strong>of</strong> renal failure between<br />
groups. Rates <strong>of</strong> RCIN 21.6% if UOP >150<br />
mL/h, 45.9% if UOP 3 mg/dL, RCIN in patients<br />
with IVF alone 4%, IVF + mannitol 24%<br />
(p=0.02), IVF + furosemide 25% (p=0.02). LOS<br />
increased by 4 days in RCIN group.<br />
SCr increased by 0.42 mg/dL +/- 0.20 treatment<br />
group vs. 0.023 mg/dL +/- 0.073 (p
Table 32.1. Studies <strong>of</strong> strategies for preventing radiocontrast-induced nephropathy<br />
(cont.)*<br />
Study Setting Study Design,<br />
Outcomes<br />
N-Acetylcysteine<br />
<strong>Patient</strong>s with SCr >1.2 mg/dL or<br />
CrCl 1.5mg/dL<br />
r<strong>and</strong>omized to pre-hydration vs.<br />
pre-hydration with low dose<br />
dopamine or aminophylline prior<br />
to cardiac catheterization. 11<br />
<strong>Patient</strong>s r<strong>and</strong>omized to prehydration<br />
+ theophylline (270 mg<br />
q am/540 mg q pm 2d before, 3d<br />
after) or placebo prior to contrast<br />
CT or DSA. 15<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
354<br />
Results<br />
RCIN developed in 2% treatment group vs. 21%<br />
control group (p=0.01). Among patients with<br />
SCr >2.5 mg/dL, RCIN 0% treatment vs. 42%<br />
controls (p=0.02)<br />
GFR reduced 85.4 +/- 3.8 mL/min controls vs.<br />
107 +/-3.6 mL/min treatment group (p 0.001).<br />
CrCl after LOCM decreased by ~18% at 24 hrs<br />
in control (p
References<br />
1. Parfrey PS, Griffiths SM, Barrett BJ, Paul MD, Genge M, Withers J, et al. Contrast<br />
material-induced renal failure in patients with diabetes mellitus, renal insufficiency, or<br />
both. A prospective controlled study. N Engl J Med. 1989;320:143-149.<br />
2. McCullough PA, Wolyn R, Rocher LL, Levin RN, O'Neill WW. Acute renal failure after<br />
coronary intervention: incidence, risk factors, <strong>and</strong> relationship to mortality. Am J Med.<br />
1997;103:368-375.<br />
3. Rich MW, Crecelius CA. Incidence, risk factors, <strong>and</strong> clinical course <strong>of</strong> acute renal<br />
insufficiency after cardiac catheterization in patients 70 years <strong>of</strong> age or older. A<br />
prospective study. Arch Intern Med. 1990;150:1237-1242.<br />
4. Levy EM, Viscoli CM, Horwitz RI. The effect <strong>of</strong> acute renal failure on mortality. A cohort<br />
analysis. JAMA. 1996;275:1489-1494.<br />
5. Lautin EM, Freeman NJ, Schoenfeld AH, Bakal CW, Haramati N, Friedman AC, et al.<br />
Radiocontrast-associated renal dysfunction: incidence <strong>and</strong> risk factors. AJR Am J<br />
Roentgenol. 1991;157:49-58.<br />
6. Brown RS, Ransil B, Clark BA. Prehydration protects against contrast nephropathy in high<br />
risk patients undergoing cardiac catheterization. J Am Soc Neprhol. 1990;1:330A.<br />
7. Barrett BJ, Carlisle EJ. Metaanalysis <strong>of</strong> the relative nephrotoxicity <strong>of</strong> high- <strong>and</strong> lowosmolality<br />
iodinated contrast media. Radiology. 1993;188:171-178.<br />
8. Solomon R, Werner C, Mann D, D'Elia J, Silva P. Effects <strong>of</strong> saline, mannitol, <strong>and</strong><br />
furosemide to prevent acute decreases in renal function induced by radiocontrast agents. N<br />
Engl J Med. 1994;331:1416-1420.<br />
9. Stevens MA, McCullough PA, Tobin KJ, Speck JP, Westveer DC, Guido-Allen DA, et al.<br />
A prospective r<strong>and</strong>omized trial <strong>of</strong> prevention measures in patients at high risk for contrast<br />
nephropathy: results <strong>of</strong> the P.R.I.N.C.E. Study. Prevention <strong>of</strong> Radiocontrast Induced<br />
Nephropathy Clinical Evaluation. J Am Coll Cardiol. 1999;33:403-411.<br />
10. Weinstein JM, Heyman S, Brezis M. Potential deleterious effect <strong>of</strong> furosemide in<br />
radiocontrast nephropathy. Nephron. 1992;62:413-415.<br />
11. Abizaid AS, Clark CE, Mintz GS, Dosa S, Popma JJ, Pichard AD, et al. Effects <strong>of</strong><br />
dopamine <strong>and</strong> aminophylline on contrast-induced acute renal failure after coronary<br />
angioplasty in patients with preexisting renal insufficiency. Am J Cardiol. 1999;83:260-<br />
263, A5.<br />
12. Kolonko A, Wiecek A, Kokot F. The nonselective adenosine antagonist theophylline does<br />
prevent renal dysfunction induced by radiographic contrast agents. J Nephrol.<br />
1998;11:151-156.<br />
13. Katholi RE, Taylor GJ, McCann WP, Woods WT, Jr., Womack KA, McCoy CD, et al.<br />
Nephrotoxicity from contrast media: attenuation with theophylline. Radiology.<br />
1995;195:17-22.<br />
14. Erley CM, Duda SH, Schlepckow S, Koehler J, Huppert PE, Strohmaier WL, et al.<br />
Adenosine antagonist theophylline prevents the reduction <strong>of</strong> glomerular filtration rate after<br />
contrast media application. Kidney Int. 1994;45:1425-1431.<br />
15. Erley CM, Duda SH, Rehfuss D, Scholtes B, Bock J, Muller C, et al. Prevention <strong>of</strong><br />
radiocontrast-media-induced nephropathy in patients with pre-existing renal insufficiency<br />
by hydration in combination with the adenosine antagonist theophylline. Nephrol Dial<br />
Transplant. 1999;14:1146-1149.<br />
355
16. Tepel M, van der Giet M, Schwarzfeld C, Laufer U, Liermann D, Zidek W. Prevention <strong>of</strong><br />
radiographic-contrast-agent-induced reductions in renal function by acetylcysteine. N Engl<br />
J Med. 2000;343:180-184.<br />
17. Koch JA, Plum J, Grabensee B, Modder U. Prostagl<strong>and</strong>in E1: a new agent for the<br />
prevention <strong>of</strong> renal dysfunction in high risk patients caused by radiocontrast media? PGE1<br />
Study Group. Nephrol Dial Transplant. 2000;15:43-49.<br />
18. Gupta RK, Kapoor A, Tewari S, Sinha N, Sharma RK. Captopril for prevention <strong>of</strong> contrastinduced<br />
nephropathy in diabetic patients: a r<strong>and</strong>omised study. Indian Heart J.<br />
1999;51:521-526.<br />
19. Seyss C, Foote EF. Calcium-channel blockers for prophylaxis <strong>of</strong> radiocontrast-associated<br />
nephrotoxicity. Ann Pharmacother. 1995;29:187-188.<br />
20. Khoury Z, Schlicht JR, Como J, Karschner JK, Shapiro AP, Mook WJ, et al. The effect <strong>of</strong><br />
prophylactic nifedipine on renal function in patients administered contrast media.<br />
Pharmacotherapy. 1995;15:59-65.<br />
21. Russo D, Testa A, Della Volpe L, Sansone G. R<strong>and</strong>omised prospective study on renal<br />
effects <strong>of</strong> two different contrast media in humans: protective role <strong>of</strong> a calcium channel<br />
blocker. Nephron. 1990;55:254-257.<br />
22. Neumayer HH, Junge W, Kufner A, Wenning A. Prevention <strong>of</strong> radiocontrast-mediainduced<br />
nephrotoxicity by the calcium channel blocker nitrendipine: a prospective<br />
r<strong>and</strong>omised clinical trial. Nephrol Dial Transplant. 1989;4:1030-1036.<br />
23. Cacoub P, Deray G, Baumelou A, Jacobs C. No evidence for protective effects <strong>of</strong><br />
nifedipine against radiocontrast-induced acute renal failure. Clin Nephrol. 1988;29:215-<br />
216.<br />
24. Cramer BC, Parfrey PS, Hutchinson TA, Baran D, Melanson DM, Ethier RE, et al. Renal<br />
function following infusion <strong>of</strong> radiologic contrast material. A prospective controlled study.<br />
Arch Intern Med. 1985;145:87-89.<br />
25. Lee JK, Warshauer DM, Bush WH, Jr., McClennan BL, Choyke PL. Determination <strong>of</strong><br />
serum creatinine level before intravenous administration <strong>of</strong> iodinated contrast medium. A<br />
survey. Invest Radiol. 1995;30:700-705.<br />
26. Taylor AJ, Hotchkiss D, Morse RW, McCabe J. PREPARED: Preparation for Angiography<br />
in Renal Dysfunction: a r<strong>and</strong>omized trial <strong>of</strong> inpatient vs outpatient hydration protocols for<br />
cardiac catheterization in mild-to-moderate renal dysfunction. Chest. 1998;114:1570-1574.<br />
27. Eisenberg RL, Bank WO, Hedgcock MW. Renal failure after major angiography. Am J<br />
Med. 1980;68:43-46.<br />
28. Teruel JL, Marcen R, Herrero JA, Felipe C, Ortuno J. An easy <strong>and</strong> effective procedure to<br />
prevent radiocontrast agent nephrotoxicity in high-risk patients. Nephron. 1989;51:282.<br />
29. Barrett BJ, Parfrey PS, Foley RN, Detsky AS. An economic analysis <strong>of</strong> strategies for the<br />
use <strong>of</strong> contrast media for diagnostic cardiac catheterization. Med Decis Making.<br />
1994;14:325-335.<br />
30. Michalson A, Franken EA, Smith W. Cost-effectiveness <strong>and</strong> safety <strong>of</strong> selective use <strong>of</strong> lowosmolality<br />
contrast media. Acad Radiol. 1994;1:59-62.<br />
31. Caro JJ, Trindade E, McGregor M. The cost-effectiveness <strong>of</strong> replacing high-osmolality<br />
with low-osmolality contrast media. AJR Am J Roentgenol. 1992;159:869-874.<br />
32. Steinberg EP, Moore RD, Powe NR, Gopalan R, David<strong>of</strong>f AJ, Litt M, et al. <strong>Safety</strong> <strong>and</strong> cost<br />
effectiveness <strong>of</strong> high-osmolality as compared with low-osmolality contrast material in<br />
patients undergoing cardiac angiography. N Engl J Med. 1992;326:425-430.<br />
356
33. Tippins RB, Torres WE, Baumgartner BR, Baumgarten DA. Are screening serum<br />
creatinine levels necessary prior to outpatient CT examinations? Radiology. 2000;216:481-<br />
484.<br />
34. Choyke PL, Cady J, DePollar SL, Austin H. Determination <strong>of</strong> serum creatinine prior to<br />
iodinated contrast media: is it necessary in all patients? Tech Urol. 1998;4:65-69.<br />
357
358
Chapter 33. Nutritional Support<br />
Neil Winawer, MD<br />
Mark V. Williams, MD<br />
Emory University School <strong>of</strong> Medicine<br />
Background<br />
There is a consensus that nutritional support should be routinely provided to intensive<br />
care unit (ICU) patients. 1 Hospitalized patients with malnutrition (macronutrient <strong>and</strong>/or<br />
micronutrient deficiency) suffer from increased infectious morbidity, prolonged hospital stays,<br />
<strong>and</strong> increased mortality. 2 Moreover, even those hospitalized medical <strong>and</strong> surgical patients<br />
without antecedent malnutrition are typically subjected to stress, infection <strong>and</strong> impaired organ<br />
function, resulting in a hypercatabolic state. Often these patients are unable to meet their caloric<br />
needs, as they are either too sick or physically unable to ingest food. Although strong evidence<br />
demonstrates that providing nutritional support for such patients results in improved clinical<br />
outcomes, the optimal method <strong>of</strong> delivery, timing <strong>of</strong> administration, <strong>and</strong> specific formulation<br />
requires further research.<br />
Practice Description<br />
There are several ways to provide nutritional support to patients in the ICU. Enteral<br />
nutrition (EN) can be administered via transoral, transnasal, or percutaneous transgastric routes,<br />
or by surgical jejunostomy. Total parental nutrition (TPN) is generally used when the enteral<br />
route is either inaccessible or its use is contraindicated. It is also used as a supplement to enteral<br />
feeding if adequate nutrition is not possible via the enteral route alone.<br />
The total caloric requirement <strong>of</strong> critically ill patients can be estimated or directly<br />
measured. Calorimetry, although accurate, is not practical in the clinical setting as it is costly,<br />
time consuming, <strong>and</strong> requires technical skill. It is also unclear that exactly matching energy input<br />
with energy expenditures improves patient outcomes. Therefore, a pragmatic approach is to<br />
attempt administration <strong>of</strong> 25 kilocalories per kilogram ideal body weight per day for most<br />
patients. 3 The total caloric daily requirement should be administered in a fluid volume consistent<br />
with the patient’s needs (usually 1mL/kcal). Protein sources should comprise 15-20% <strong>of</strong> the total<br />
daily calorie requirement. The generally accepted amount <strong>of</strong> protein is between 1.2 <strong>and</strong> 1.5 g/kg<br />
per day, except in severe losses such as burns. Glucose should comprise 30-70% <strong>of</strong> the total<br />
calories <strong>and</strong> fats 15-30%. 1<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Malnutrition in hospitalized patients <strong>of</strong>ten goes unrecognized. 4,5 Early studies reported a<br />
prevalence <strong>of</strong> malnutrition in 30-50% <strong>of</strong> hospitalized patients. 6 A later study revealed that up to<br />
40% <strong>of</strong> patients were malnourished at the time <strong>of</strong> their admission. 7 The majority <strong>of</strong> these patients<br />
continued to be nutritionally depleted throughout their hospital course. These patients are also at<br />
a greater risk for the development <strong>of</strong> severe malnutrition than those patients whose nutritional<br />
status was adequate at the time <strong>of</strong> admission.<br />
Unfortunately, there is no single, readily available measure <strong>of</strong> malnutrition that is both<br />
sensitive <strong>and</strong> specific in critically ill patients. 1 Most studies have used body mass index<br />
(BMI=weight (kg)/height (m)) <strong>and</strong>/or anthropometry (measuring skin fold thickness) to assess<br />
patients’ nutritional status. 7 BMI alone is not a sensitive indicator <strong>of</strong> protein-energy malnutrition<br />
as it does not distinguish between depletion <strong>of</strong> fat or muscle. 4 In a large number <strong>of</strong> studies,<br />
359
malnutrition has been defined as a BMI ≤20 kg/m <strong>and</strong> a triceps skin fold thickness (TSF) or midarm<br />
muscle circumference (MAMC)
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Nutritional supplementation in hospitalized patients may reduce mortality <strong>and</strong> is<br />
associated with weight gain (see Table 33.1). 13 However, there are no r<strong>and</strong>omized controlled<br />
trials comparing supplemental nutrition to starvation in critically ill patients. Research does show<br />
that patients not receiving any nutritional support for more than 2 weeks postoperatively have a<br />
much higher complication <strong>and</strong> mortality rate than patients receiving TPN or some short-term<br />
glucose administration. 14 A large body <strong>of</strong> research in the past 2 decades has focused on<br />
determining the ideal type <strong>and</strong> method <strong>of</strong> delivery <strong>of</strong> nutritional support.<br />
A meta-analysis comparing supplemental TPN to st<strong>and</strong>ard care (oral diet as tolerated <strong>and</strong><br />
intravenous dextrose) found no effect on mortality (relative risk (RR) 1.03, 95% CI: 0.81-1.31). 11<br />
There was a trend toward a lower complication rate among those receiving TPN (RR 0.84, 95%<br />
CI: 0.64-1.09), but this is due mainly to benefit among malnourished patients. 11 There are no<br />
data from r<strong>and</strong>omized controlled trials to support the use <strong>of</strong> supplemental TPN among patients<br />
with an intact gastrointestinal tract (“If the gut works, use it”).<br />
Several studies have evaluated the use <strong>of</strong> supplemental EN in surgical patients. These<br />
studies <strong>of</strong>ten used surrogate outcomes, but one r<strong>and</strong>omized double-blind trial <strong>of</strong> early EN versus<br />
st<strong>and</strong>ard diet as tolerated following surgery found fewer total complications (26.7% vs. 63.3%,<br />
p=0.009), fewer infectious complications (6.7% vs. 46.7%, p
10%) in those patients receiving TPN. 6 This benefit appears to be due entirely to significant<br />
reduction in surgical morbidity among patients who are severely malnourished. 22 Therefore,<br />
preoperative TPN may be <strong>of</strong> benefit in severely malnourished patients undergoing major<br />
gastrointestinal surgery, but EN should be used instead, if possible. The use <strong>of</strong> postoperative<br />
TPN has been evaluated in 8 prospective r<strong>and</strong>omized trials <strong>of</strong> patients undergoing<br />
gastrointestinal surgery. <strong>Patient</strong>s in the combined TPN group experienced an increased rate <strong>of</strong><br />
complications (27.3% vs. 16.0%; p
developing major infectious complications (odss ratio 0.47, 95% CI: 0.32-0.70), <strong>and</strong> hospital<br />
length <strong>of</strong> stay (reduced 2.5 days, 95% CI: 1.0-4.0). 29 Though there was no “significant”<br />
difference in mortality between patients receiving immune-enhanced versus st<strong>and</strong>ard nutritional<br />
support, there was a trend toward increased mortality (odds ratio 1.77, 95% CI: 1.00-3.12). 29<br />
Potential for Harm<br />
TPN has been associated with an increase in septic complications. <strong>Patient</strong>s may also be<br />
placed at increased risk from attempts to obtain central venous access. While EN decreases the<br />
overall rate <strong>of</strong> infectious complications when compared to TPN, it may place patients, especially<br />
those with head injuries, at greater risk <strong>of</strong> aspiration pneumonia. Use <strong>of</strong> a promotility agent, such<br />
as metoclopramide, has reduced this risk in one study. 30<br />
Costs <strong>and</strong> Implementation<br />
TPN is itself expensive. Moreover, hospital costs can climb dramatically if patients who<br />
receive TPN suffer infectious complications that prolong their hospital stay. Obtaining central<br />
venous access also increases the costs <strong>of</strong> administration. Nutritional support teams (NSTs) can<br />
help ensure that patients are receiving adequate nutrition while at the same time reducing the rate<br />
<strong>of</strong> line complications. 28 However, the use <strong>of</strong> NSTs is expensive <strong>and</strong> has not been subject to a<br />
cost analysis. 28 The economics <strong>of</strong> early postoperative enteral nutrition was studied in a recent<br />
nonr<strong>and</strong>omized, prospective, clinical trial <strong>of</strong> patients undergoing bowel resections. There was a<br />
reduction in variable cost <strong>of</strong> $1531 per success in the treatment group (p=0.02) <strong>and</strong> $4450 total<br />
cost savings per success in the treatment group (p=0.04). 31 Immunonutrition is significantly more<br />
expensive than st<strong>and</strong>ard enteral formulas but in 2 prospective r<strong>and</strong>omized controlled trials <strong>of</strong><br />
patients undergoing upper gastrointestinal surgery it has been shown to be cost-effective. In one<br />
study, the cost per patient for nutrition <strong>and</strong> complications in the control group was $1633<br />
compared to $783 in the immunonutrition group. 32 In another study the cost per patient for<br />
nutrition <strong>and</strong> complications in the control group was $1107 compared with $755 in the treatment<br />
group. 33<br />
Comment<br />
The literature on appropriate nutritional support for critically ill patients is complex to<br />
analyze. A wide variety <strong>of</strong> treatment modalities have been evaluated in a fairly heterogeneous<br />
patient population, making interpretation <strong>of</strong> the various studies difficult.<br />
Nonetheless, there are several general recommendations that can be made. First,<br />
malnutrition leads to poor outcomes <strong>and</strong> should be avoided or treated if present. <strong>Patient</strong>s who<br />
have prolonged starvation for more than 2 weeks are at a significantly increased risk for<br />
complications. EN in these patients decreases this risk, <strong>and</strong> should be administered to all<br />
critically ill medical <strong>and</strong> surgical patients who can tolerate it. There are no data to support the<br />
use <strong>of</strong> TPN in critically ill patients who have a functional gastrointestinal tract. However, TPN<br />
may be <strong>of</strong> benefit preoperatively when given electively to severely malnourished patients<br />
undergoing gastrointestinal resection. Postoperative administration <strong>of</strong> TPN worsens outcomes.<br />
There is no evidence to suggest that postoperative enteral nutrition is superior to advancing a<br />
patient’s oral diet, as tolerated, if they are capable <strong>of</strong> eating, but research into early<br />
administration <strong>of</strong> EN via the jejunum is ongoing.<br />
Use <strong>of</strong> immune-enhanced EN appears to significantly reduce infectious complications,<br />
number <strong>of</strong> days <strong>of</strong> mechanical ventilation, <strong>and</strong> overall hospital length <strong>of</strong> stay, though it does not<br />
appear to affect overall mortality (<strong>and</strong> may increase it). Of note, study groups with increased<br />
363
mortality rates may also have a diminished incidence <strong>of</strong> infectious complications <strong>and</strong> decreased<br />
length <strong>of</strong> stay artifactually due to patients dying early in the study. One expert interpreted the<br />
published research as strongly suggesting that immune-enhancing formulas should be used in<br />
critically ill patients, especially surgical patients, despite their high cost. 34 Nonetheless, a large<br />
multicenter RCT is needed to resolve with certainty whether or not use <strong>of</strong> these expensive<br />
formulas is safe <strong>and</strong> improves outcomes in critically ill patients before formally recommending<br />
their widespread use. 29<br />
364
Table 33.1. Studies evaluating nutritional support*<br />
Study Study Design,<br />
Outcomes<br />
Routine protein energy<br />
supplementation in<br />
adults; systematic<br />
review 13<br />
TPN in critically ill<br />
patients; metaanalysis<br />
11<br />
TPN in surgical<br />
patients; metaanalysis<br />
12<br />
Immunonutrition in<br />
critically ill patients;<br />
systematic review <strong>and</strong><br />
meta-analysis 28<br />
Immunonutrition in<br />
patients with critical<br />
illness <strong>and</strong> cancer;<br />
meta-analysis 29<br />
EN vs. TPN in<br />
seriously ill<br />
hospitalized patients;<br />
systematic review <strong>and</strong><br />
meta-analysis 2<br />
Enteral<br />
immunonutrition in the<br />
critically ill; doubleblind<br />
r<strong>and</strong>omized<br />
controlled trial 26<br />
Level 1A, Level<br />
1<br />
Level 1A, Level<br />
1<br />
Level 1A, Level<br />
1<br />
Level 1A, Level<br />
1<br />
Level 1A, Level<br />
1<br />
Level 1A, Level<br />
1<br />
Level 1, Level<br />
1<br />
Results (95% Confidence Interval)<br />
Reduced mortality: OR 0.66 (0.48 to 0.91)<br />
Increased body weight gain (%):<br />
Oral Sip Feeds: 2.39 (1.80 to 2.96)<br />
Oral natural feeds: 5.36 (1.73 to 8.99)<br />
Nasogastric feeding: 4.04 (3.15 to 4.94)<br />
Percutaneous or enteral feeding, enterostomy:<br />
-1.38 (-2.35 to -0.41)<br />
No effect on mortality: RR 1.03 (0.81 to 1.31)<br />
Complication rate: RR 0.84 (0.64 to 1.09)<br />
Complication rate (malnourished subgroup):<br />
RR 0.52 (0.30 to 0.91)<br />
No effect on mortality: RR 0.97 (0.76 to 1.24)<br />
No effect on mortality (malnourished subgroup):<br />
RR 1.13 (0.75 to 1.71)<br />
Major complication rate (malnourished subgroup):<br />
RR 0.52 (0.30 to 0.91)<br />
No effect on mortality: RR 1.05 (0.78 to 1.41)<br />
Reduction in infection rate: RR 0.67 (0.50 to 0.89)<br />
Reduction in days <strong>of</strong> mechanical ventilation:<br />
2.6 days (0.1 to 5.1)<br />
Reduction in hospital LOS: 2.9 days (1.4 to 4.4)<br />
Trend toward increased mortality:<br />
OR 1.77 (1.00 to 3.12)<br />
Reduction in infectious complications:<br />
OR 0.47 (0.32 to 0.70)<br />
No reduction in pneumonia risk: OR 0.91 (0.53 to 1.56)<br />
Reduction in hospital LOS: 2.5 days (4.0 to 1.0)<br />
Reduction in septic complications:<br />
EN 18% vs. TPN 35% (p=0.01)<br />
Reduction in infectious complications:<br />
EN 16% vs. TPN 35% (p=0.01)<br />
Overall analysis (immunonutrition vs. control formula)<br />
Mortality: 48% vs. 44% (p=0.36)<br />
Total days <strong>of</strong> mechanical ventilation (median):<br />
4 vs. 4 days (p=NS)<br />
Hospital LOS (median): 12 vs. 13 days (p=NS)<br />
Early enteral feeding subgroup<br />
Total days <strong>of</strong> mechanical ventilation (median):<br />
6 vs. 10.5 days (p=0.007)<br />
Hospital LOS (median): 15.5 vs. 20 days (p=0.03)<br />
* LOS indicates length <strong>of</strong> stay; NS, not statistically significant; OR, odds ratio; <strong>and</strong> RR, relative risk.<br />
365
References<br />
1. Cerra FB, Benitez MR, Blackburn GL, Irwin RS, Jeejeebhoy K, Katz DP, et al. Applied<br />
nutrition in ICU patients. A consensus statement <strong>of</strong> the American College <strong>of</strong> Chest<br />
Physicians. Chest. 1997;111:769-778.<br />
2. Heyl<strong>and</strong> DK. Enteral <strong>and</strong> parenteral nutrition in the seriously ill, hospitalized patient: A<br />
critical review <strong>of</strong> the evidence. Journ <strong>of</strong> Nutr Health & Aging. 2000;1:31-41.<br />
3. Jolliet P, Pichard C, Biolo G, Chiolero R, Grimble G, Leverve X, et al. Enteral nutrition in<br />
intensive care patients: a practical approach. Intensive Care Med. 1998;24:848-859.<br />
4. McWhirter JP, Pennington CR. The incidence <strong>and</strong> recognition <strong>of</strong> malnutrition in hospital.<br />
BMJ. 1994;308:945-948.<br />
5. Potter JM, Klipstein K, Reilly JJ, Roberts MA. The nutritional status <strong>and</strong> clinical course <strong>of</strong><br />
acute admissions to a geriatric unit. Age Ageing. 1995;24:131-136.<br />
6. Torosian MH. Perioperative nutrition support for patients undergoing gastrointestinal<br />
surgery: critical analysis <strong>and</strong> recommendations. World J Surg. 1999;23:565-569.<br />
7. Beattie AH, Prach AT, Baxter JP, Pennington CR. A r<strong>and</strong>omized controlled trial evaluating<br />
the use <strong>of</strong> enteral nutritional supplements postoperatively in malnourished surgical patients.<br />
Gut. 2000;46:813-818.<br />
8. Adam S, Batson S. A study <strong>of</strong> problems associated with the delivery <strong>of</strong> enteral feed in<br />
critically ill patients in five ICUs in the UK. Intensive Care Med. 1997;23:261-266.<br />
9. McClave SA, Sexton LK, Spain DA. Enteral tube feeding in the intensive care unit: factors<br />
impeding adequate delivery. Crit Care Med. 1999;27:1252-1256.<br />
10. DeJonghe B, Appere-De-Vechi C, Fournier M, Tran B, Merrer J, Melchior JC, et al. A<br />
prospective survey <strong>of</strong> nutritional support practices in intensive care unit patients: What is<br />
prescribed? What is delivered? Crit Care Med. 2001;29:8-12.<br />
11. Heyl<strong>and</strong> DK, MacDonald S, Keefe L, Drover JW. Total parental nutrition in the critically<br />
ill patient. A meta-analysis. JAMA. 1998;280:2013-2019.<br />
12. Heyl<strong>and</strong> DK, Montalvo M, MacDonald S, Keefe L, Su XY, Drover JW. Total parenteral<br />
nutrition in the surgical patient: a meta-analysis. Can J Surg. 2001;44:102-111.<br />
13. Potter J, Langhorne P, Roberts M. Routine protein energy supplementation in adults:<br />
systematic review. BMJ. 1998;317:495-501.<br />
14. S<strong>and</strong>strom R, Drott C, Hylt<strong>and</strong>er A. The effect <strong>of</strong> postoperative intravenous feeding (TPN)<br />
on outcome following major surgery evaluated in a r<strong>and</strong>omized study. Ann Surg.<br />
1993;217:185-195.<br />
15. Beier-Holgersen R, Boesby S. Influence <strong>of</strong> postoperative enteral nutrition on postsurgical<br />
infections. Gut. 1996;39:833-835.<br />
16. Singh G, Ram RP, Khanna SK. Early postoperative enteral feeding in patients with<br />
nontraumatic intestinal perforation <strong>and</strong> peritonitis. J Am Coll Surg. 1998;187:142-146.<br />
17. Kalfarentzos F, Kehagias J, MEad N, Kokkinis K, Gogos CA. Enteral nutrition is superior<br />
to parenteral nutrition in severe acute pancreatitis: Results <strong>of</strong> a r<strong>and</strong>omized prospective<br />
trial. Br J Surg. 1997;84:1665-1669.<br />
18. Iovinelli G, Marsili I, Varassi. Nutritional support after total laryngectomy. JPEN.<br />
1993;17:445-448.<br />
366
19. Moore FA, Moore EE, Jones TN, McCroskey BL, Petersen VM. TEN versus TPN<br />
following major abdominal trauma reduced septic morbidity. J <strong>of</strong> Trauma. 1989;29:916-<br />
923.<br />
20. Moore FA, Feliciano DV, Andrassy RJ, McArdle AH, Booth FV, Morgenstein-Wagner<br />
TB, et al. Early enteral feeding, compared with parental, reduces postoperative septic<br />
complications. The results <strong>of</strong> a meta-analysis. Ann Surg. 1992;216:172-183.<br />
21. Young B, Ott L, Twyman. The effect <strong>of</strong> nutritional support on outcome from severe head<br />
injury. J Neurosurg. 1987;67:668-676.<br />
22. The Veterans Affairs Total Parenteral Nutrition Cooperative Study Group. Perioperative<br />
total parenteral nutrition in surgical patients. N Engl J Med. 1991;325:525-532.<br />
23. Zaloga GP, Bortenschlager L, Black KW, Prielipp R. Immediate postoperative enteral<br />
feeding decreases weight loss <strong>and</strong> improves healing after abdominal surgery in rats. Crit<br />
Care Med. 1992;20:115-119.<br />
24. Mochizuki H, Trocki O, Dominioni L, Brackett KA, J<strong>of</strong>fe SN, Alex<strong>and</strong>er JW. Mechanism<br />
<strong>of</strong> prevention <strong>of</strong> postburn hypermetabolism <strong>and</strong> catabolism by early enteral feeding. Ann<br />
Surg. 1984;200:297-310.<br />
25. Moore EE, Jones TN. Benefits <strong>of</strong> immediate jejunostomy feeding after major abdominal<br />
trauma - a prospetive r<strong>and</strong>omized study. J <strong>of</strong> Trauma. 1986;26:874-880.<br />
26. Atkinson S, Sieffert E, Bihari D. A prospective, r<strong>and</strong>omized, double-blind, controlled<br />
clinical trial <strong>of</strong> enteral immunonutrition in the critically ill. Crit Care Med. 1998;26:1164-<br />
1172.<br />
27. Heyl<strong>and</strong> DK, Drover J. Does immunonutrition make an impact? It depends on the analysis.<br />
Crit Care Med. 2000;28:906.<br />
28. Beale RJ, Bryg DJ, Bihari DJ. Immunonutrition in the critically ill: A systematic review <strong>of</strong><br />
clinical outcome. Crit Care Med. 1999;27:2799-2805.<br />
29. Heys SD, Walker LG, Smith I, Eremin O. Enteral nutritional supplementation with key<br />
nutrients in patients with critical illness <strong>and</strong> cancer. A meta-analyis <strong>of</strong> r<strong>and</strong>omized<br />
controlled clinical trials. Ann Surg. 1999;229:467-477.<br />
30. Jooste CA, Mustoe J, Collee G. Metoclopraamide improves gastric motility in critically ill<br />
patients. Intensive Care Med. 1999;25:464-468.<br />
31. Hedberg A, Lairson DR, Aday LA, Chow J, Suki R, Houston S, et al. Economic<br />
implications <strong>of</strong> an early postoperative enteral feeding protocol. J Am Diet Assoc.<br />
1999;99:802-807.<br />
32. Senkal M, Zumtobel V, Bauer KH, Marpe B, Wolfram G, Frei A, et al. Outcome <strong>and</strong> costeffectiveness<br />
<strong>of</strong> perioperative enteral immunonutrition in patients undergoing elective<br />
upper gastrointestinal tract surgery. Arch Surg. 1994;134:1309-1316.<br />
33. Senkal M, Mumme A, Eickh<strong>of</strong>f U, Geier B, Spath G, Wulfert D, et al. Early postoperative<br />
enteral immunonutrition: Clinical outcome <strong>and</strong> cost-comparison analysis in surgical<br />
patients. Crit Care Med. 1997;25:1489-1496.<br />
34. Zaloga GP. Immune-enhancing diets: Where's the beef? Crit Care Med. 1998;26:1143-<br />
1146.<br />
367
Chapter 34. Prevention <strong>of</strong> Clinically Significant Gastrointestinal Bleeding in<br />
Intensive Care Unit <strong>Patient</strong>s<br />
Daniel D. Dressler, MD<br />
Mark V. Williams, MD<br />
Kimberly Rask, MD, PhD<br />
Emory University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Stress-related gastric ulceration was first described in the early 1970s, 1 <strong>and</strong> has since<br />
received extensive study. Appreciation <strong>of</strong> the physiologic changes that promote stress-related<br />
gastritis 2-4 <strong>and</strong> general improvements in the care <strong>of</strong> critically ill patients have likely played a role<br />
in reducing the frequency <strong>of</strong> this complication. 3 Nonetheless, the use <strong>of</strong> specific pharmacologic<br />
agents for the prevention <strong>of</strong> stress-related gastrointestinal (GI) bleeding is increasingly promoted<br />
as st<strong>and</strong>ard therapy in the ICU setting. 5-8 Despite the common use <strong>of</strong> these agents, controversy<br />
about the evidence supporting this practice remains. 5,9 Because specific pharmacologic<br />
prophylaxis for stress ulceration may increase the risk <strong>of</strong> other complications (eg, nosocomial<br />
pneumonia 10 ) <strong>and</strong> is associated with significant costs, 11 recent efforts have focused on<br />
delineating an appropriate definition <strong>of</strong> clinically important GI bleeding <strong>and</strong> identifying patients<br />
who derive a clear benefit from pharmacologic prevention <strong>of</strong> this complication.<br />
Practice Description<br />
This chapter reviews evidence supporting the use <strong>of</strong> pharmacologic therapies for stress<br />
ulcer <strong>and</strong> GI bleeding prophylaxis in the ICU setting. We considered the use <strong>of</strong> histamine-2<br />
receptor blocking agents (H2-receptor antagonists) <strong>and</strong> sucralfate, a mucosal protecting agent.<br />
Although the efficacy <strong>of</strong> aluminum- <strong>and</strong> magnesium-based antacids has been demonstrated, their<br />
use is limited by dosing frequency <strong>and</strong> side effects. 12 Proton-pump inhibitors (PPIs) <strong>and</strong> enteral<br />
nutrition have also shown benefit in small studies, but there are yet no large r<strong>and</strong>omized<br />
evaluations. 6,12,13<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The risk <strong>of</strong> stress ulceration <strong>and</strong> subsequent GI bleeding depends on a patient’s<br />
underlying illness, its severity, <strong>and</strong> related comorbidities. Using liberal criteria for assessment <strong>of</strong><br />
bleeding, early reports estimated the incidence <strong>of</strong> GI bleeding due to stress ulceration in ICU<br />
patients to be as high as 5-25%. 14,15 However, more recent prospective studies, using stricter<br />
definitions <strong>of</strong> clinically significant GI bleeding <strong>and</strong> following large cohorts <strong>of</strong> patients, have<br />
revealed more modest estimates <strong>of</strong> 0.1% in low-risk ICU patients <strong>and</strong> 2.8% in ventilated<br />
patients. 2,13,16<br />
Opportunities for Impact<br />
A recent, prospectively collected database <strong>of</strong> over 7000 patients admitted to surgical <strong>and</strong><br />
medical intensive care units at a tertiary care center (1988-95) revealed that 59.9% <strong>of</strong> patients<br />
received pharmacologic prophylaxis with ranitidine or sucralfate for stress ulceration. 16 Another<br />
study assessed the annual change in stress ulcer prophylaxis in nearly 3000 patients admitted to a<br />
medical intensive care unit at a tertiary care hospital from 1993 to 1996. It found a significant<br />
decrease in rates <strong>of</strong> prophylaxis, with a 71% rate in 1993 progressively decreasing to a 21% rate<br />
in 1996, likely reflecting emerging evidence regarding high <strong>and</strong> low-risk populations in the<br />
368
literature during that time. 17 Identifying appropriate patients for stress ulcer prophylaxis <strong>and</strong><br />
avoiding its use in other cases both <strong>of</strong>fer potential opportunities for improving patient safety in<br />
the ICU.<br />
Study Designs<br />
Although multiple meta-analyses have reviewed the prophylaxis <strong>of</strong> GI bleeding in<br />
critically ill patients, many were performed more than a decade ago. 18-20 Since then, results <strong>of</strong><br />
additional r<strong>and</strong>omized trials have been published. More recent, well-designed search strategies,<br />
strict quality-scoring <strong>of</strong> literature, <strong>and</strong> rigorous adherence to criteria defining clinically<br />
significant GI bleeding have improved the breadth <strong>and</strong> quality <strong>of</strong> literature captured, likely<br />
making older systematic reviews less relevant. Because <strong>of</strong> these changes in quality measures as<br />
well as overlap <strong>of</strong> information <strong>and</strong> trials from older reviews contained within more recent ones,<br />
only those meta-analyses published within the last 5 years (<strong>and</strong> large r<strong>and</strong>omized controlled<br />
trials not included in these reviews) are included here.<br />
A 1996 meta-analysis found 269 studies through a broad search <strong>of</strong> multiple databases <strong>and</strong><br />
extensive efforts to find unpublished trials. 5 Sixty-three <strong>of</strong> these met inclusion criteria. The<br />
analysis evaluated multiple types <strong>of</strong> prophylaxis for GI bleeding, including antacids, H2antagonists,<br />
<strong>and</strong> sucralfate. The study authors compared each type <strong>of</strong> prophylaxis to placebo <strong>and</strong><br />
to each other when literature was available. The same study group then followed-up unanswered<br />
questions from their meta-analysis with a large r<strong>and</strong>omized controlled trial (RCT) <strong>of</strong> 1200<br />
patients, comparing ranitidine to sucralfate without a placebo group. 21<br />
A more recent meta-analysis abstracted evidence from the literature to compare ranitidine<br />
(instead <strong>of</strong> all H2-antagonists) versus placebo, <strong>and</strong> sucralfate versus placebo for GI bleed<br />
prophylaxis in the ICU. 9 The search methods were less rigorous <strong>and</strong> the target population<br />
narrower than the 1996 meta-analysis, but the authors claimed to assess a more clinically<br />
relevant population <strong>and</strong> outcome.<br />
Results from 2 prospective cohort studies that evaluated risk factors for occurrence <strong>of</strong><br />
clinically important GI bleeding also provide key recommendations regarding clinical<br />
practice. 2,13<br />
Study Outcomes<br />
Both meta-analyses <strong>and</strong> the large RCT reported outcomes <strong>of</strong> clinically important GI<br />
bleeding <strong>and</strong> nosocomial pneumonia (Level 1). The 1996 meta-analysis <strong>and</strong> r<strong>and</strong>omized trial<br />
also evaluated overt GI bleeding <strong>and</strong> mortality. Cohort studies (Level 2) present results <strong>of</strong><br />
multivariate regression analyses, identifying the salient risk factors for clinically important GI<br />
bleeding in the ICU patient.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Table 34.1 summarizes the outcomes for the 2 meta-analyses <strong>and</strong> large RCT. The<br />
disparity in results <strong>of</strong> the meta-analyses is partly due to differing article selection criteria. While<br />
the 1996 analysis evaluated all H2-antagonists, the 2000 study excluded trials evaluating<br />
cimetidine because <strong>of</strong> its virtual replacement (attributable to its undesirable side effect pr<strong>of</strong>ile)<br />
by ranitidine in current clinical practice. 5,9 Excluding studies with cimetidine substantially<br />
reduced the number <strong>of</strong> studies available for review. Additionally, the 1996 study evaluated<br />
comparisons <strong>of</strong> therapies to each other as well as to placebo, while the 2000 study only<br />
considered trials <strong>of</strong> effectiveness with a placebo control as relevant.<br />
369
The 2000 meta-analysis found no statistically significant reduction in clinically important<br />
GI bleeding when comparing H2-antagonist agents with placebo, or sucralfate to placebo. 9<br />
However, the 1996 study revealed a significant difference between ranitidine <strong>and</strong> placebo for<br />
clinically important GI bleeding. 5 It was unable to demonstrate such a difference between<br />
ranitidine <strong>and</strong> sucralfate. The results <strong>of</strong> these 2 reviews reveal discrepant findings, in large part<br />
due to exclusion <strong>of</strong> cimetidine in the 2000 study. However, that study also included at least one<br />
moderate-sized trial that used remarkably strict criteria for defining clinically important GI<br />
bleeding. This factor likely contributed an element <strong>of</strong> bias towards the null result in that metaanalysis.<br />
Furthermore, the large RCT (1998) comparing ranitidine to sucralfate, without a placebo<br />
comparison group, revealed a statistically significant difference in the rate <strong>of</strong> clinically important<br />
upper GI bleeding, favoring H2-antagonists. 21 In this RCT the number needed to treat (NNT)<br />
with ranitidine compared to placebo to prevent one clinically important GI bleed compared to<br />
placebo was 47. However, no reduction in mortality or length <strong>of</strong> ICU stay was found.<br />
Interpretation <strong>of</strong> the results <strong>of</strong> the RCT are complicated by: 1) wide confidence intervals<br />
surrounding the relative risk estimate; 2) very small numbers <strong>of</strong> patients with clinically<br />
important GI bleeding (ranitidine group: 10/596; sucralfate group: 23/604), only 42% <strong>of</strong> whom<br />
had endoscopic confirmation <strong>of</strong> the source <strong>of</strong> bleed; 3) a large number <strong>of</strong> patients (70%)<br />
receiving concomitant enteral nutrition, believed to reduce the risk <strong>of</strong> GI bleeding (see Chapter<br />
33); <strong>and</strong> 4) unreported incidence <strong>of</strong> coagulopathy or duration <strong>of</strong> prophylaxis prior to GI<br />
bleeding. 22<br />
The 2 large cohort studies found respiratory failure (odds ratio 15.6, p
Costs <strong>and</strong> Implementation<br />
One cost-effectiveness analysis extracted relevant literature from MEDLINE between<br />
1985 <strong>and</strong> 1995, representing 15 controlled clinical trials (Level 1). 19 Assumptions made in the<br />
analysis included equal efficacy for the H2-antagonist cimetidine <strong>and</strong> sucralfate, a baseline risk<br />
<strong>of</strong> bleeding <strong>of</strong> 6%, <strong>and</strong> a 50% risk-reduction due to prophylaxis. The average cost per bleeding<br />
episode averted was $1144 for sucralfate, but 6.5 times greater for cimetidine. However, low-risk<br />
patients amassed a cost per bleeding episode averted <strong>of</strong> $103,725 compared to a cost <strong>of</strong> $279 for<br />
very high-risk patients. Cost per bleeding episode averted increased significantly if the risk <strong>of</strong><br />
nosocomial pneumonia was included in the analysis.<br />
The large RCT reported no difference in the duration <strong>of</strong> ICU stay for the group treated<br />
with ranitidine compared with the group treated with sucralfate (9 days for each group). 21<br />
However, no placebo group was used.<br />
Comment<br />
Overall, the evidence available in the literature does not conclusively demonstrate that<br />
the benefits <strong>of</strong> GI prophylaxis outweigh its risks for every patient admitted to the intensive care<br />
unit. As refinements have been made in defining clinically important <strong>and</strong> significant episodes <strong>of</strong><br />
GI bleeding, the population that may reap benefit from prophylaxis has narrowed substantially.<br />
In turn, a reduction in the use <strong>of</strong> prophylactic agents has followed. Most recent estimates reveal a<br />
negligible incidence <strong>of</strong> clinically important stress-related GI bleeding in low-risk ICU patients<br />
(approximately 0.1%) <strong>and</strong> a small incidence (less than 3%) in higher-risk patients. Improvements<br />
in overall ICU care <strong>and</strong> use <strong>of</strong> enteral nutrition may be contributing to this decrease in incidence.<br />
Although a statistical benefit <strong>of</strong> H2-antagonists compared with placebo or with sucralfate<br />
has been shown in some studies, the overall clinical benefit <strong>of</strong> these agents has been disputed.<br />
Some research shows a greater absolute number <strong>of</strong> nosocomial pneumonia cases related to<br />
therapy with H2-antagonists than the benefit from reduction in GI bleeding, causing the NNH to<br />
be potentially smaller than the NNT (see Subchapter 17.4). Furthermore, the overall cost-tobenefit<br />
ratio <strong>of</strong> prophylaxis increases dramatically in lower-risk patients, as shown in the<br />
analysis noted above that accounted for increases in nosocomial pneumonia. 21 Thus, the clear<br />
benefit <strong>of</strong> administering H2-antagonists to prevent GI bleeding among many patients in an ICU<br />
remains to be shown. This is partly due to variations in the definition <strong>of</strong> clinically significant<br />
bleeding <strong>and</strong> pneumonia. Additional large trials are necessary to identify the patients who truly<br />
derive net benefit from this practice, as published evidence does not yet support this practice for<br />
many patients currently receiving the therapy.<br />
At the present time, clinicians may consider use <strong>of</strong> prophylactic agents, an H2-antagonist<br />
or sucralfate, to prevent clinically important GI bleeding in very high-risk patients admitted to<br />
the ICU. Such patients may include those with respiratory failure, coagulopathy, renal failure,<br />
<strong>and</strong>/or burns (the latter group has been excluded from most studies because its risk is believed to<br />
be so great). However, the risk <strong>of</strong> pneumonia may influence clinicians to use prophylactic agents<br />
only in patients with multiple risk factors for GI bleeding, <strong>and</strong> simply provide enteral nutrition to<br />
others at less risk. The use <strong>of</strong> PPIs requires further study before any recommendation can be<br />
made regarding them. Further research should focus on identifying subgroups <strong>of</strong> patients who do<br />
derive net benefit from ranitidine or other acid-reducing agents.<br />
371
Table 34.1. Studies evaluating effectiveness <strong>of</strong> pharmacologic prophylaxis <strong>of</strong> ICU patients<br />
to prevent clinically significant GI bleeding*<br />
EFFECTIVENESS<br />
Study Study Design, Outcomes Comparison Groups Effect Size (95% CI)<br />
Cook,<br />
1996 5<br />
Messori,<br />
2000 9<br />
Cook,<br />
1998 21<br />
Level 1A, Level 1 H2-antagonist vs. placebo<br />
Sucralfate vs. placebo<br />
H2-antagonist vs. sucralfate<br />
372<br />
OR 0.44 (0.22-0.88)<br />
OR 1.49 (0.42-5.27)<br />
OR 1.28 (0.27-6.11)<br />
Level 1A, Level 1 Ranitidine vs. placebo OR 0.72 (0.30-1.70)<br />
Level 1, Level 1 Ranitidine vs. sucralfate RR 0.44 (0.21-0.92)<br />
NNT 47<br />
* CI indicates confidence interval; NNT, number needed to treat (for benefit); OR, odds ratio;<br />
<strong>and</strong> RR, relative risk.<br />
Table 34.2. Studies evaluating harm (nosocomial pneumonia) due to pharmacologic<br />
prophylaxis <strong>of</strong> ICU patients to prevent GI bleeding*<br />
Study Study Design, Outcomes<br />
Measured<br />
Cook,<br />
1996 5<br />
Messori,<br />
2000 9<br />
Cook,<br />
1998 21<br />
HARM<br />
Comparison Groups Effect Size (95% CI)<br />
Level 1A, Level 1 Sucralfate vs. H2-antagonist OR 0.78 (0.60-1.01)<br />
Level 1A, Level 1 Ranitidine vs. sucralfate OR 1.35 (1.07-1.70)<br />
NNH 21<br />
Level 1, Level 1 Ranitidine vs. sucralfate RR 1.18 (0.92-1.51)<br />
NNH 34<br />
* CI indicates confidence interval; NNH, number needed to harm; OR, odds ratio; <strong>and</strong> RR,<br />
relative risk.<br />
References<br />
1. Lucas C, Sugawa C, Riddle J, Rector F, Rosenberg B, Walt A. Natural History <strong>and</strong> surgical<br />
dilemma <strong>of</strong> "stress" gastric bleeding. Arch Surgery. 1971;102:266-273.<br />
2. Cook D, Fuller H, Guyatt G, Marshall J, Leasa D, Hall R, et al. Risk factors for<br />
gastrointestinal bleeding in critically ill patients. N Engl J Med. 1994;330:377-381.<br />
3. Haglund U. Stress ulcers. Sc<strong>and</strong> J Gastroenterol. 1990;175:27-33.<br />
4. Shoemaker W, Bl<strong>and</strong> R, PL A. Therapy <strong>of</strong> critically ill postoperative patients based on<br />
outcome prediction <strong>and</strong> prospective clinical trials. Surg Clin North Am. 1985;65:811-833.<br />
5. Cook D, Reeve B, Guyatt G, Heyl<strong>and</strong> D, Griffith L, Buckingham L, et al. Stress ulcer<br />
prophylaxis in critically ill patients: resolving discordant meta-analyses. JAMA.<br />
1996;275:308-314.
6. Beejay U, Wolfe M. Acute gastroinestinal bleeding in the intensive care unit: the<br />
gastroenterologist's perspective. Gastroenterol Clin North Am. 2000;29:309-336.<br />
7. Friedman S, Manaker S, Weinhouse G. Stress ulcer prophylaxis in the intensive care unit.<br />
In: Rose BD, ed. UpToDate. Wellesley, MA: UpToDate, Inc.; 2001.<br />
8. ASHP Commission on Therapeutics. Therapeutic guidelines on stress ulcer prophylaxis.<br />
Am J Health Syst Pharm. 1999;56:347-379.<br />
9. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A. Bleeding <strong>and</strong> pneumonia in<br />
intensive care patients given ranitidine <strong>and</strong> sucralfate for prevention <strong>of</strong> stress ulcer: metaanalysis<br />
<strong>of</strong> r<strong>and</strong>omised controlled trials. BMJ. 2000; 21:1-7.<br />
10. Ben-Menachem T, McCarthy B, Fogel R, Schiffman R, Patel R, Zarowitz B, et al.<br />
Prophylaxis for stress-related gastrointestinal hemmorhage: a cost effectiveness analysis.<br />
Crit Care Med. 1996;24:338-345.<br />
11. Cook D. Stress ulcer prophylaxis gastrointestinal bleeding <strong>and</strong> nosocomial pneumonia.<br />
Best evidence synthesis. Sc<strong>and</strong> J Gastroenterol. 1995;210:48-52.<br />
12. Hastings P, Skillman J, Bushnell L, Silen W. Antacid titration in the prevention <strong>of</strong> acute<br />
gastrointestinal bleeding: a controlled, r<strong>and</strong>omized trial in 100 critically ill patients. N Engl<br />
J Med. 1978;298:1041-1045.<br />
13. Raff T, Germann G, Hartmann B. The value <strong>of</strong> early enteral nutrition in the prophylaxis <strong>of</strong><br />
stress ulceration in the severely burned patient. Burns. 1997;23:313-318.<br />
14. Ben-menachem T, Fogel R, Patel R, Touchette M, Zarowitz B, Hadzijahic N, et al.<br />
Prophylaxis for stress-related gastric hemorrhage in the medical intensive care unit: a<br />
r<strong>and</strong>omized, controlled, single-blind study. Ann Intern Med. 1994; 21:568-575.<br />
15. Shuman R, Schuster D, Zuckerman G. Prophylactic therapy for stress ulcer bleeding: a<br />
reappraisal. Ann Int Med. 1987;106:562-567.<br />
16. Pimentel M, Roberts D, Bernstein C, Hoppensack M, Duerksen D. Clinically significant<br />
gastrointestinal bleeding in critically ill patients in an era <strong>of</strong> prophylaxis. Am J<br />
Gastroenterol. 2000;95:2801-2806.<br />
17. Devlin J, Ben-Menachem T, Ulep S, Peters M, Fogel R, Zarowitz B. Stress ulcer<br />
prophylaxis in medical ICU <strong>Patient</strong>s: annual utilization in relation to the incidence <strong>of</strong><br />
endoscopically proven stress ulceration. Ann Pharmacother. 1998;32:869-874.<br />
18. Tryba M. Prophylaxis <strong>of</strong> stress ulcer bleeding: a meta-analysis. J Clin Gastroenterol.<br />
1991;3(Suppl. 2):S44-S55.<br />
19. Lacroix J, Infante-Rivard C, Jenicek M, Gauthier M. Prophylaxis <strong>of</strong> upper gastrointestinal<br />
bleeding in intensive care units: a meta analysis. Crit Care Med 1989;17:862-869.<br />
20. Cook D, Witt L, Cook R, Guyatt G. Stress ulcer prophylaxis in the critically ill: a meta<br />
analysis. Am J Med. 1991;91:519-527.<br />
21. Cook D, Guyatt G, Marshall J, Leasa D, Fuller H, Hall R, et al. A comparison <strong>of</strong> sucralfate<br />
<strong>and</strong> ranitidine for the prevention <strong>of</strong> upper gastrointestinal bleeding in patients requiring<br />
mechanical ventilation. N Engl J Med. 1998;338:791-797.<br />
22. Hinds C, Fletcher S. Ranitidine reduced clinically important gastrointestinal bleeding in<br />
patients who required mechanical ventilation. Gut. 1999;44:10-11.<br />
23. Burgess P, Larson G, Davidson P, Brown J, Metz C. Effect <strong>of</strong> ranitidine on intragastric pH<br />
<strong>and</strong> stress-related upper gastrointestinal bleeding in patients with severe head injury. Dig<br />
Dis Sci. 1995;40:645-650.<br />
24. Cook D, Heyl<strong>and</strong> D, Griffith L, Cook R, Marshall J, Pagliarello J. Risk factors for<br />
clinically important upper gastrointestinal bleeding in patients requiring mechanical<br />
ventilation. Crit Care Med. 1999;27:2812-2817.<br />
373
374
375
Chapter 35. Reducing Errors in the Interpretation <strong>of</strong> Plain Radiographs <strong>and</strong><br />
Computed Tomography Scans<br />
Sunil Kripalani, MD<br />
Mark V. Williams, MD<br />
Kimberly Rask, MD, PhD<br />
Emory University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong> Health<br />
Background<br />
Misinterpretation <strong>of</strong> radiographic studies is a common source <strong>of</strong> medical error in both the<br />
inpatient <strong>and</strong> outpatient arenas. 1,2 Of particular concern is the significant number <strong>of</strong><br />
misinterpretations <strong>of</strong> plain radiographs <strong>and</strong> cranial computed tomography (CT) scans in<br />
emergency departments (ED) or urgent care settings by non-radiologists. 3-6 The prevalence <strong>of</strong><br />
this patient safety issue may result from the large volume <strong>of</strong> patients receiving these radiologic<br />
tests, which are <strong>of</strong>ten done outside normal working hours when radiologists are not available to<br />
provide an initial interpretation. This chapter focuses on practices to reduce non-radiologists’<br />
higher rates <strong>of</strong> misinterpretation <strong>of</strong> these commonly ordered studies.<br />
Intuitively, it would seems that institutions could minimize the number <strong>of</strong> such mistakes<br />
by routinely having studies interpreted by the most accurate <strong>and</strong> experienced physicians, usually<br />
radiologists. The American College <strong>of</strong> Radiology (ACR) recommends that all imaging<br />
procedures culminate in a written, expert opinion from a radiologist or another licensed<br />
physician specifically trained in diagnostic radiology. 7,8 However, due to the associated costs,<br />
fewer than 20% <strong>of</strong> hospitals have full-time on-site coverage by a board-certified radiologist. 9<br />
Instead, radiologists are generally available for 8 to 12 hours a day <strong>and</strong> may not provide an<br />
interpretation until the following morning, particularly when studies are performed after normal<br />
working hours. 9 For many routine examinations, it may be possible to delay interpretation<br />
without harm to patients. If results are needed urgently <strong>and</strong> a radiologist is unavailable, other<br />
physicians (eg, emergency physicians, hospitalists, neurologists) must take responsibility for the<br />
initial interpretation.<br />
<strong>Patient</strong> safety may be enhanced by improving the diagnostic accuracy <strong>of</strong> these physicians<br />
or by implementing other systems to prevent initial misinterpretations from adversely affecting<br />
patient care. Several strategies reviewed here that may be effective in reducing these errors<br />
include educational courses to improve the diagnostic accuracy <strong>of</strong> non-radiologists, on-site<br />
coverage by radiology residents, <strong>and</strong> m<strong>and</strong>atory subsequent reinterpretation <strong>of</strong> studies by<br />
radiologists.<br />
Another approach that deserves mention is to have the initial readings made by <strong>of</strong>f-site<br />
radiologists or other specialists using a teleradiology link. Teleradiology has been effective in<br />
facilitating emergent neurosurgical consultation prior to the interhospital transfer <strong>of</strong> patients with<br />
head injury. 10-12 Teleradiology also allows rural physicians to obtain remote consults for selected<br />
patients, 13-16 which in one report led to treatment changes in 26% <strong>of</strong> cases. 17 Despite<br />
teleradiology’s impact in these two specific settings, few studies have tested its accuracy <strong>and</strong><br />
utility in more general circumstances. 18 Teleradiology <strong>of</strong> course requires the use <strong>of</strong> digitized<br />
rather than film-based radiographs. In two mixed case series, discrepancies <strong>of</strong> interpretation<br />
between digitized <strong>and</strong> original radiographs occurred in approximately 10% <strong>of</strong> cases, 13,14 with<br />
significant discrepancies in 1.5-5%. 19,20 For more subtle findings the sensitivity <strong>of</strong> on-screen<br />
images may be as low as 49%, 21 leading several investigators to conclude that teleradiology is<br />
376
inferior to film interpretation for difficult cases. 21-23 At present, it appears that image quality is<br />
the major reason behind the variable performance <strong>of</strong> teleradiology. Although the ACR has<br />
established detailed st<strong>and</strong>ards for equipment <strong>and</strong> image resolution, 24 radiology practices <strong>of</strong>ten<br />
utilize less expensive alternatives for viewing images, including personal computers. 23,25,26 The<br />
variation in practice, rapid evolution <strong>of</strong> technology, <strong>and</strong> lack <strong>of</strong> large prospective trials make it<br />
difficult to examine teleradiology from the st<strong>and</strong>point <strong>of</strong> patient safety. Therefore, this chapter<br />
will not discuss teleradiology in detail; reviews can be found elsewhere. 27-30<br />
Interventions to improve the technical quality <strong>of</strong> imaging studies, such as ensuring proper<br />
patient positioning <strong>and</strong> film exposure, are certainly important but are also outside the scope <strong>of</strong><br />
this chapter. Finally, although radiologists make mistakes in interpreting films, 31-37 this chapter<br />
focuses on practices to reduce the higher rate <strong>of</strong> misinterpretations made by non-radiologists.<br />
Practice Descriptions<br />
Training Courses<br />
Training courses for non-radiologists are presumably common, but we could locate only<br />
2 descriptions in the literature. 38,39 Both courses concerned interpretation <strong>of</strong> cranial CT scans,<br />
were conducted by radiologists, <strong>and</strong> targeted ED residents <strong>and</strong> faculty physicians. The training<br />
lasted 1-2 hours <strong>and</strong> consisted <strong>of</strong> a review <strong>of</strong> neuroanatomy <strong>and</strong> a small library <strong>of</strong> CT scans. Two<br />
continuous quality improvement initiatives were also recently described, in which radiologists<br />
provided regular feedback to ED physicians about their radiograph interpretations. 40,41<br />
Initial Interpretations by Radiology Residents<br />
Initial interpretation <strong>of</strong> ED films by radiology residents follows no st<strong>and</strong>ardized practice.<br />
Their hours <strong>of</strong> coverage <strong>and</strong> the degree to which they interact with ED physicians vary widely. 42<br />
It is also unclear to what extent emergency physicians rely on the resident interpretations when<br />
they are available.<br />
Review <strong>of</strong> All Studies by Radiologists<br />
Reinterpretation <strong>of</strong> all studies by a radiologist is already commonly used. When a<br />
radiologist finds that a study was initially misinterpreted, the medical error has already occurred.<br />
In this case, the safety practice concerns how the radiologist communicates the corrected<br />
interpretation to providers in order to minimize the risk <strong>of</strong> harm to patients. The method <strong>of</strong><br />
communication varies among health care facilities <strong>and</strong> includes placing the interpretation in the<br />
patient’s medical record, sending a report to the referring physician, or contacting the physician<br />
or patient directly for more urgent concerns (see also Subchapter 42.4). 7<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Many investigators have focused on the prevalence <strong>of</strong> all ED readings that are discordant<br />
with radiologists’ subsequent interpretations (with the radiologists’ interpretation assumed to be<br />
the “gold st<strong>and</strong>ard”). Against this st<strong>and</strong>ard, ED physicians <strong>and</strong> residents misinterpret 1-16% <strong>of</strong><br />
plain radiographs 43-51 <strong>and</strong> approximately 35% <strong>of</strong> cranial CT scans. 52 However, many discordant<br />
readings involve subtle or incidental findings that are not clinically significant (ie, they do not<br />
affect patient management or outcome). From the st<strong>and</strong>point <strong>of</strong> patient safety, it is the rate <strong>of</strong><br />
clinically significant misinterpretations <strong>and</strong> related errors in management that are <strong>of</strong> concern.<br />
Most ED studies show that important errors in interpretation occur in 1-3% <strong>of</strong> plain<br />
radiographs. 40,43,46-51 However, one pediatric study calculated a 6.8% rate <strong>of</strong> significant<br />
377
adiograph misinterpretation. 53 The most common error is failure to recognize an extremity<br />
fracture. 43,46,51,54 Lufkin <strong>and</strong> colleagues examined the readings <strong>of</strong> 16,410 consecutive radiographs<br />
<strong>and</strong> the effect <strong>of</strong> emergency physicians’ confidence on the accuracy <strong>of</strong> their interpretations. 55<br />
When ED physicians were confident in their interpretation, the rate <strong>of</strong> discordant readings was<br />
1.2%. More importantly, the rate <strong>of</strong> clinically significant errors was only 0.1%.<br />
Rates <strong>of</strong> misinterpretation are even higher for CT scans. In one study, ED residents <strong>and</strong><br />
attending physicians overlooked new infarcts, mass lesions, <strong>and</strong> cerebral edema, as well as<br />
parenchymal, subarachnoid, <strong>and</strong> subdural hemorrhages. 52 The rate <strong>of</strong> clinically important errors<br />
was 20-25%, with failure to recognize a cerebral infarction the most common. 52 It is unclear how<br />
many patients had adverse outcomes as a result, but the authors estimated that less than 1% <strong>of</strong><br />
patients were managed inappropriately by ED staff. 52 Other studies confirm that the radiographic<br />
accuracy <strong>of</strong> non-radiologists in detecting a major cerebral infarction is poor, with<br />
misinterpretation rates <strong>of</strong> 30-40%. 56-59 This may directly impact patient outcomes since<br />
management <strong>of</strong> suspected stroke, specifically determining eligibility for thrombolytic therapy,<br />
requires an immediate <strong>and</strong> accurate interpretation <strong>of</strong> the CT. In addition, intraobserver <strong>and</strong><br />
interobserver reliability are fair to poor, with kappa (κ) values ranging from 0.41-0.20. 57,59<br />
Radiology residents have better accuracy in interpretation <strong>of</strong> cranial CT scans, though not<br />
as high as that <strong>of</strong> certified radiologists. As part <strong>of</strong> a departmental quality control program, Lal<br />
<strong>and</strong> colleagues found the rate <strong>of</strong> significant misinterpretations by radiology residents was 0.9%. 60<br />
Roszler et al reported moderate or major errors in 2.1% <strong>of</strong> resident interpretations <strong>of</strong> posttraumatic<br />
head CT scans. 61 Wysoki <strong>and</strong> colleagues 62 found that major discrepancies between the<br />
interpretations <strong>of</strong> residents <strong>and</strong> staff radiologists occurred in 1.7% <strong>of</strong> neuroradiologic studies <strong>and</strong><br />
that the rate was significantly higher when CT scans were abnormal rather than normal (12.2%<br />
vs. 1.5%). Residents missed 9% <strong>of</strong> intracranial hemorrhages <strong>and</strong> 17% <strong>of</strong> cranial or facial<br />
fractures.<br />
Opportunities for Impact<br />
The accepted st<strong>and</strong>ard <strong>of</strong> care, supported by the ACR, calls for routine review <strong>of</strong> all<br />
radiologic studies by a radiologist or other qualified physician in a timely fashion. 8 Given the<br />
available data on staffing patterns, 9 we estimate that on-site radiologists are available to interpret<br />
radiographs about half the time. Radiologists generally read any remaining studies the following<br />
day. In academic centers, the percentage <strong>of</strong> films initially interpreted by residents may range<br />
from 20% to 100%, <strong>and</strong> the availability <strong>of</strong> on-call residents is highly variable. 42 It is also unclear<br />
what proportion <strong>of</strong> academic <strong>and</strong> community hospitals provides radiographic training courses for<br />
non-radiologists.<br />
Study Designs<br />
Few studies specifically focus on methods to reduce clinically significant<br />
misinterpretations <strong>of</strong> radiographs <strong>and</strong> CT scans. No r<strong>and</strong>omized trials have evaluated the<br />
effectiveness <strong>of</strong> the 3 patient safety practices identified above. Table 35.1 shows 4 prospective<br />
before-after studies (Level 2 design) <strong>of</strong> educational interventions <strong>and</strong> quality improvement<br />
initiatives. 38-41 Evidence for the other 2 strategies (initial interpretations by radiology residents<br />
<strong>and</strong> review <strong>of</strong> all studies by radiologists) is limited to descriptive studies reporting rates <strong>of</strong><br />
discordant interpretations for various groups <strong>of</strong> physicians (see above).<br />
378
Study Outcomes<br />
Most studies reported the rate <strong>of</strong> clinically significant misinterpretations (Level 2<br />
outcome). 38,40,41,53,55,63 However, the definition <strong>of</strong> “clinically significant” was subjective <strong>and</strong><br />
varied among reports. One trial reported only the percentage <strong>of</strong> correct CT interpretations (Level<br />
3), without describing the significance <strong>of</strong> specific errors. 39<br />
Evidence for Effectiveness <strong>of</strong> the <strong>Practices</strong><br />
Levitt <strong>and</strong> colleagues 38 found a significant improvement in cranial CT scan<br />
interpretations by ED physicians after a one-hour training course. However, there were several<br />
limitations <strong>of</strong> this study. The intervention was not tested before implementation, raising concerns<br />
regarding its reliability <strong>and</strong> reproducibility. Previously published research at the same institution<br />
showed a 38.7% misinterpretation rate <strong>of</strong> cranial CT scans by ED physicians. 39 Because<br />
physicians were aware <strong>of</strong> these results, they may have engaged in their own efforts that led to<br />
improvement in CT scan interpretation. The improvement might also be explained by significant<br />
changes in case-mix, which was not measured. A multicenter study by Perron et al 39 showed that<br />
a reproducible educational course significantly improved residents’ ability to interpret CT scans.<br />
However, the study is limited by possible selection bias (participation in the course was<br />
voluntary; post-test data were obtained in only 61 <strong>of</strong> 83 subjects). Although subjects in the study<br />
by Perron were retested after a 3-month washout period, neither course has been shown to result<br />
in sustained improvement. Quality improvement programs such as those described by Espinosa<br />
<strong>and</strong> Preston 40,41 successfully reduced the rate <strong>of</strong> misinterpreted radiographs <strong>and</strong> number <strong>of</strong><br />
callbacks, respectively. Through ongoing feedback <strong>and</strong> review <strong>of</strong> misinterpreted radiographs,<br />
Espinosa <strong>and</strong> colleagues lowered the rate <strong>of</strong> important errors from 3% to 0.3%, a relative risk<br />
reduction <strong>of</strong> 90%. The initiative described by Preston led to a 42.9% relative reduction in<br />
callbacks to the ED, although the absolute magnitude was less impressive (0.3% absolute risk<br />
reduction).<br />
We found no studies that compared the readings <strong>of</strong> ED physicians with those <strong>of</strong><br />
radiology residents in a real-world environment. Only one investigation compared their<br />
diagnostic skill in an experimental situation. 18 Eng <strong>and</strong> colleagues exposed 4 groups <strong>of</strong><br />
physicians (ED attendings, ED residents, radiology attendings, <strong>and</strong> radiology residents) to a<br />
series <strong>of</strong> 120 radiographs <strong>and</strong> calculated their receiver operating characteristic (ROC) curves.<br />
The radiographs were selected for their difficulty, <strong>and</strong> ED physicians had previously<br />
misinterpreted many <strong>of</strong> them. The area under the ROC curve was 0.15 higher for radiology<br />
faculty than for ED attendings (95% CI: 0.10-0.20) <strong>and</strong> 0.07 higher for all faculty than all<br />
residents (95% CI: 0.02-0.12). Compared with ED faculty, radiology residents had an additional<br />
area under the ROC curve <strong>of</strong> 0.08 (95% CI: 0.02-0.14).<br />
379
Potential for Harm<br />
There is a potential for both radiologists 51 <strong>and</strong> non-radiologists 64 to overread films (ie,<br />
falsely identify non-existent findings), which may result in unnecessary diagnostic testing or<br />
treatment. Since the vast majority <strong>of</strong> this literature considers the radiologist’s reading to be the<br />
gold st<strong>and</strong>ard, the proportion <strong>of</strong> discrepancies in interpretation that are due to false-positive<br />
readings by radiologists is unclear.<br />
Costs <strong>and</strong> Implementation<br />
Only one study reported the costs <strong>of</strong> false-positive readings by ED physicians, which<br />
averaged $85 per false-positive radiograph. 64 Given the paucity <strong>of</strong> research, it is not possible to<br />
estimate the costs <strong>of</strong> implementing universal review <strong>of</strong> emergency films by an on-site<br />
radiologist. Finally, no studies have measured the costs <strong>of</strong> misreads to the patient <strong>and</strong> health care<br />
system (eg, transportation, return visits, repeat or unnecessary studies, unnecessary medications)<br />
which could potentially <strong>of</strong>fset staffing costs. The marginal costs <strong>of</strong> after-hours staffing alone<br />
may be prohibitive, especially for low-volume sites. No information is available to estimate the<br />
cost <strong>of</strong> establishing a high-quality teleradiology program, or <strong>of</strong> educational programs for ED<br />
physicians or trainees.<br />
Comment<br />
The rates <strong>of</strong> misinterpreted radiographs <strong>and</strong> CT scans are high in many studies. Of<br />
particular concern is the 20-25% rate <strong>of</strong> clinically significant errors in reading cranial CT scans,<br />
even among experienced emergency physicians <strong>and</strong> neurologists. Although plain films are<br />
correctly interpreted more than 90% <strong>of</strong> the time, even “low” error rates <strong>of</strong> 1% or less are<br />
important given the sheer number <strong>of</strong> films.<br />
The relative roles <strong>of</strong> ED physicians, radiology residents, certified radiologists, <strong>and</strong> other<br />
physicians will depend on their accuracy <strong>and</strong> their cost. The literature is notable for the relative<br />
dearth <strong>of</strong> studies <strong>of</strong> interventions to reduce radiologic misinterpretation by non-radiologists, even<br />
though dozens <strong>of</strong> studies have documented the problem. Where radiographs must be interpreted<br />
by non-radiologists, there is limited evidence that brief educational interventions <strong>and</strong> continuous<br />
quality improvement programs improve diagnostic accuracy. (Chapter 54 reviews general issues<br />
surrounding changing practice behavior through education.) It is possible that radiologists’<br />
routine review <strong>of</strong> plain radiographs may not be cost-effective when the non-radiologist clinician<br />
has a high level <strong>of</strong> confidence in the initial interpretation, but this has not been rigorously<br />
established. Coverage by radiology residents may add back-up accuracy to the reading <strong>of</strong> an ED<br />
attending, but Eng’s study was biased toward a difficult set <strong>of</strong> radiographs that had previously<br />
been misinterpreted by emergency physicians. 18 The added value <strong>of</strong> a radiology resident’s<br />
interpretation may be much lower in actual practice.<br />
One avenue <strong>of</strong> research that could yield effective safety practices is human resource<br />
management. Essentially, staffing could be optimized such that specific tests could be triaged to<br />
the individual with the highest diagnostic accuracy. This kind <strong>of</strong> intervention would require<br />
assessment <strong>of</strong> both potential coverage gaps <strong>and</strong> the skill mix <strong>of</strong> the available labor force. A<br />
formal protocol might be developed <strong>and</strong> tested to identify coverage vulnerabilities <strong>and</strong> develop<br />
realistic coverage options based on measured performance <strong>of</strong> the available personnel at a<br />
particular site. While the value <strong>of</strong> this strategy is entirely speculative, it draws on general<br />
management science <strong>and</strong> could be explored more explicitly in health care.<br />
380
Table 35.1. Educational interventions for non-radiologists*<br />
Study Study Setting Intervention Study Design,<br />
Outcomes<br />
Levitt,<br />
1997 38<br />
Espinosa,<br />
2000 41<br />
Preston,<br />
1998 40<br />
Perron,<br />
1998 39<br />
14 ED physicians at a<br />
level 2 trauma center<br />
in California<br />
ED physicians at an<br />
academic hospital in<br />
New Jersey, 1993-99<br />
ED physicians <strong>and</strong><br />
radiologists at a 150bed<br />
community<br />
hospital in Louisiana,<br />
1990-95<br />
83 ED residents at 5<br />
academic centers in<br />
the southeast US,<br />
1997-98<br />
One-hour course on<br />
cranial CT interpretation<br />
Training on plain<br />
radiograph interpretation<br />
<strong>and</strong> ongoing feedback<br />
from radiologists about<br />
their errors<br />
Continuous quality<br />
improvement initiative<br />
with regular review <strong>of</strong><br />
film discrepancies<br />
Two-hour course on<br />
neuroanatomy <strong>and</strong> cranial<br />
CT interpretation<br />
381<br />
Level 2,<br />
Level 2<br />
Level 2,<br />
Level 2<br />
Level 2,<br />
Level 2<br />
Level 2,<br />
Level 3<br />
Results†<br />
Clinically significant<br />
misinterpretations<br />
decreased from 23.6% to<br />
4.0%<br />
(ARR 19.6%, RRR 83%)<br />
Clinically significant false<br />
negative interpretations:<br />
from 3% to 0.3%<br />
(ARR 2.7%, RRR 90%)<br />
<strong>Patient</strong> callbacks to ED for<br />
clinically significant<br />
misinterpretation: from<br />
0.7% to 0.4%<br />
(ARR 0.3%, RRR 42.9%)<br />
Correct interpretation <strong>of</strong> a<br />
series <strong>of</strong> 12 CT scans:<br />
baseline 60%; 3 months<br />
after the course, 78%<br />
(p
10. Goh KY, Tsang KY, Poon WS. Does teleradiology improve inter-hospital management <strong>of</strong><br />
head injury? Can J Neurol Sci.1997;24:235-239.<br />
11. Goh KY, Lam CK, Poon WS. The impact <strong>of</strong> teleradiology on the inter-hospital transfer <strong>of</strong><br />
neurosurgical patients. British Journal <strong>of</strong> Neurosurgery 1997;11:52-56.<br />
12. Gray WP, Somers J, Buckley TF. Report <strong>of</strong> a national neurosurgical emergency<br />
teleconsulting system. Neurosurgery 1998;42:103-108.<br />
13. Franken EA, Berbaum KS, Smith WL, Chang PJ, Owen DA, Bergus GR. Teleradiology for<br />
rural hospitals: analysis <strong>of</strong> a field study. J Telemed Telecare 1995;1:202-208.<br />
14. Franken EA, Berbaum KS, Br<strong>and</strong>ser EA, D'Aless<strong>and</strong>ro MP, Schweiger GD, Smith ML.<br />
Pediatric radiology at a rural hospital: value <strong>of</strong> teleradiology <strong>and</strong> subspecialty consultation.<br />
AJR 1997;168:1349-1352.<br />
15. Franken EA, Harkens KL, Berbaum KS. Teleradiology consultation for a rural hospital:<br />
patterns <strong>of</strong> use. Acad Radiol 1997;4:492-496.<br />
16. Lambrecht CJ. Emergency physicians' roles in a clinical telemedicine network. Ann Emerg<br />
Med 1997;30:670-674.<br />
17. Lee JK, Renner JB, Saunders BF, Stamford PP, Bickford TR, Johnston RE, et al. Effect <strong>of</strong><br />
real-time teleradiology on the practice <strong>of</strong> the emergency department physician in a rural<br />
setting: initial experience. Acad Radiol 1998;5:533-538.<br />
18. Eng J, Mysko WK, Weller GER, et al. Interpretation <strong>of</strong> emergency department<br />
radiographs: a comparison <strong>of</strong> emergency medicine physicians with radiologists, residents<br />
with faculty, <strong>and</strong> film with digital display. AJR 2000;175:1233-1238.<br />
19. Gale ME, Vincent ME, Robbins AH. Teleradiology for remote diagnosis: a prospective<br />
multi-year evaluation. J Digit Imaging 1997;10:47-50.<br />
20. DeCorato DR, Kagetsu NJ, Ablow RC. Off-hours interpretation <strong>of</strong> radiologic images <strong>of</strong><br />
patients admitted to the emergency department: efficacy <strong>of</strong> teleradiology. AJR<br />
1995;165:1293-1296.<br />
21. Scott WW, Rosenbaum JE, Ackerman SJ, Reichle RL, Magid D, Weller JC, et al. Subtle<br />
orthopedic fractures: teleradiology workstation versus film interpretation. Radiology<br />
1993;187:811-815.<br />
22. Scott WW, Bluemke DA, Mysko WK, Weller GE, Kelen GD, Reichle RK, et al.<br />
Interpretation <strong>of</strong> emergency department radiographs by radiologists <strong>and</strong> emergency<br />
medicine physicians: teleradiology workstation versus radiograph readings. Radiology<br />
1995;195:223-229.<br />
23. Calder LD, Maclean JR, Bayliss AP, Gilbert FJ, Grant AM. The diagnostic performance <strong>of</strong><br />
a PC-based teleradiology link. Clin Radiol 1999;54:659-664.<br />
24. American College <strong>of</strong> Radiology. ACR st<strong>and</strong>ard for teleradiology; 1998.<br />
25. O'Reilly S, Spedding R, Dearden C, Loane M. Can x-rays be accurately interpreted using a<br />
low-cost telemedicine system? Emerg Med J 1998;15:312-314.<br />
26. Berry RF, Barry MH. Evaluation <strong>of</strong> a personal-computer-based teleradiology system<br />
serving an isolated Canadian community. Can Assoc Radiol J 1998;49:7-11.<br />
27. Goldberg MA. Teleradiology <strong>and</strong> telemedicine. Radiol Clin North Am. 1996;34:647-665.<br />
28. Bol<strong>and</strong> GWL. Teleradiology: another revolution in radiology? Clin Radiol 1998;53:547-<br />
553.<br />
29. Baker SR, Festa S. The use <strong>of</strong> teleradiology in an emergency setting with speculation on<br />
the near future. Radiol Clin North Am 1999;37:1035-1044.<br />
30. Benger J. A review <strong>of</strong> telemedicine in accident <strong>and</strong> emergency: the story so far. J Accid<br />
Emerg Med 2000;17:157-164.<br />
382
31. Herman PG, Gerson DE, Hessel SJ, Mayer BS, Watnick M, Blesser B, et al. Disagreements<br />
in chest roentgen interpretation. Chest 1975;68:278-282.<br />
32. Herman PG, Hessel SJ. Accuracy <strong>and</strong> its relationship to experience in the interpretation <strong>of</strong><br />
chest radiographs. Invest Radiol 1975;10:62-67.<br />
33. Hillman BJ, Swensson RG, Hessel SJ, Herson DE, Herman PG. The value <strong>of</strong> consultation<br />
among radiologists. Am J Roentgenol 1976;127:807-809.<br />
34. Hillman BJ, Hessel SJ, Swensson RG, Herman PG. Improving diagnostic accuracy: a<br />
comparison <strong>of</strong> interactive <strong>and</strong> Delphi consultations. Invest Radiol 1977;12:112-115.<br />
35. Hessel SJ, Herman PG, Swensson RG. Improving performance by multiple interpretations<br />
<strong>of</strong> chest radiographs: effectiveness <strong>and</strong> cost. Radiology 1978;127:589-594.<br />
36. Rhea JT, Potsaid MS, DeLuca SA. Errors <strong>of</strong> interpretation as elicited by a quality audit <strong>of</strong><br />
an emergency radiology facility. Radiology 1979;132:277-280.<br />
37. Markus JB, Somers S, O'Malley BP, Stevenson GW. Double-contrast barium enema<br />
studies: effect <strong>of</strong> multiple reading on perception error. Radiology 1990;175:155-6.<br />
38. Levitt MA, Dawkins R, Williams V, Bullock S. Abbreviated educational session improves<br />
cranial computed tomography scan interpretations be emergency physicians. Ann Emerg<br />
Med 1997;30:616-621.<br />
39. Perron AD, Huff JS, Ullrich CG, Heafner MD, Kline JA. A multicenter study to improve<br />
emergency medicine residents' recognition <strong>of</strong> intracranial emergencies on computed<br />
tomography. Ann Emerg Med 1998;32:554-562.<br />
40. Preston CA, Marr JJ, Amaraneni KK, Suthar BS. Reduction <strong>of</strong> "call-backs" to the<br />
emergency department due to discrepancies in the plain radiograph interpretation. Am J<br />
Emerg Med 1998;16:160-162.<br />
41. Espinosa JA, Nolan TW. Reducing errors made by emergency physicians in interpreting<br />
radiographs: longitudinal study. BMJ 2000;320:737-740.<br />
42. Townsend RR, Manco-Johnson ML. Night call in U.S. radiology residency programs. Acad<br />
Radiol 1995;2:810-815.<br />
43. Warren JS, Lara K, Connor PD, Cantrell J, Hahn RG. Correlation <strong>of</strong> emergency department<br />
radiographs: results <strong>of</strong> a quality assurance review in an urban community hospital setting. J<br />
Am Board Fam Pract 1993;6:255-259.<br />
44. Snow DA. Clinical significance <strong>of</strong> discrepancies in roentgenographic film interpretation in<br />
an acute walk-in area. J Gen Intern Med 1986;1:295-299.<br />
45. Robinson PJA, Wilson D, Coral A, Murphy A, Verow P. Variation between experienced<br />
observers in the interpretation <strong>of</strong> accident <strong>and</strong> emergency radiographs. Br J Radiol<br />
1999;72:323-330.<br />
46. Thomas HG, Mason AC, Smith RM, Fergusson CM. Value <strong>of</strong> radiograph audit in an<br />
accident service department. Injury 1992;23:47-50.<br />
47. Gratton MC, Salomone JA, Watson WA. Clinically significant radiograph<br />
misinterpretations at an emergency medicine residency program. Ann Emerg Med<br />
1990;19:497-502.<br />
48. Walsh-Kelly CM, Melzer-Lange MD, Hennes HM, Lye P, Hegenbarth M, Sty J, et al.<br />
Clinical impact <strong>of</strong> radiograph misinterpretation in a pediatric ED <strong>and</strong> the effect <strong>of</strong><br />
physician training level. Am J Emerg Med 1995;13:262-264.<br />
49. Klein EJ, Koenig M, Diekema DS, Winters W. Discordant radiograph interpretation<br />
between emergency physicians <strong>and</strong> radiologists in a pediatric emergency department.<br />
Pediatr Emerg Care 1999;15:245-248.<br />
383
50. Fleisher G, Ludwig S, McSorley M. Interpretation <strong>of</strong> pediatric x-ray films by emergency<br />
department pediatricians. Ann Emerg Med 1983;12:153-158.<br />
51. Brunswick JE, Ilkhanipour K, Seaberg DC, McGill L. Radiographic interpretation in the<br />
emergency department. Am J Emerg Med 1996;14:346-348.<br />
52. Alfaro D, Levitt MA, English DK, Williams V, Eisenberg R. Accuracy <strong>of</strong> interpretation <strong>of</strong><br />
cranial computed tomography scans in an emergency medicine residency program. Ann<br />
Emerg Med 1995;25:169-174.<br />
53. Simon HK, Khan NS, Nordenberg DF, Wright JA. Pediatric emergency physician<br />
interpretation <strong>of</strong> plain radiographs: is routine review by a radiologist necessary <strong>and</strong> costeffective?<br />
Ann Emerg Med 1996;27:295-298.<br />
54. Freed HA, Shields NN. Most frequently overlooked radiographically apparent fractures in<br />
a teaching hospital emergency department. Ann Emerg Med 1984;31:900-904.<br />
55. Lufkin KC, Smith SW, Matticks CA, Brunette DD. Radiologists' review <strong>of</strong> radiographs<br />
interpreted confidently by emergency physicians infrequently leads to changes in patient<br />
management. Ann Emerg Med 1998;31:202-207.<br />
56. Wardlaw JM, Dorman PJ, Lewis SC, S<strong>and</strong>ercock PAG. Can stroke physicians <strong>and</strong><br />
neuroradiologists identify signs <strong>of</strong> early cerebral infarction on CT? J Neurol Neurosurg<br />
Psychiatry 1999;67:651-653.<br />
57. Grotta JC, Chiu D, Lu M, et al. Agreement <strong>and</strong> variability in the interpretation <strong>of</strong> early CT<br />
changes in stroke patients qualifying for intravenous rtPA therapy. Stroke 1999;30:1528-<br />
1533.<br />
58. Kalafut MA, Schriger DL, Saver JL, Starkman S. Detection <strong>of</strong> early CT signs <strong>of</strong> >1/3<br />
middle cerebral artery infarctions : interrater reliability <strong>and</strong> sensitivity <strong>of</strong> CT interpretation<br />
by physicians involved in acute stroke care. Stroke 2000;31:1667-1671.<br />
59. Schriger DL, Kalafut M, Starkman S, Krueger M, Saver JL. Cranial computed tomography<br />
interpretation in acute stroke: physician accuracy in determining eligibility for<br />
thrombolytic therapy. JAMA 1998;279:1293-1297.<br />
60. Lal NR, Murray UM, Eldevik OP, Desmond JS. Clinical consequences <strong>of</strong><br />
misinterpretations <strong>of</strong> neuroradiologic CT scans by on-call radiology residents. AJNR<br />
2000;21:124-129.<br />
61. Roszler MH, McCarroll KA, Rashid T, Donovan KR, Kling GA. Resident interpretation <strong>of</strong><br />
emergency computed tomographic scans. Invest Radiol 1991;26:374-376.<br />
62. Wysoki MG, Nassar CJ, Koenigsberg RA, Novelline RA, Faro SH, Faerber EN. Head<br />
trauma: CT scan interpretation by radiology residents versus staff radiologists. Radiology<br />
1998;208:125-128.<br />
63. Seltzer SE, Hessel SJ, Herman PG, Swensson RG, Sheriff CR. Resident film<br />
interpretations <strong>and</strong> staff review. AJR 1980;137:129-133.<br />
64. Walsh-Kelly CM, Hennes HM, Melzer-Lange MD. False-positive preliminary radiograph<br />
interpretations in a pediatric emergency department: clinical <strong>and</strong> economic impact. Am J<br />
Emerg Med 1997;15:354-356.<br />
384
Chapter 36. Pneumococcal Vaccination Prior to Hospital Discharge<br />
Scott Fl<strong>and</strong>ers, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Streptococcus pneumoniae (pneumococcus) is a leading cause <strong>of</strong> serious communityacquired<br />
infections, especially pneumonia, the sixth leading cause <strong>of</strong> death in the United States. 1<br />
It causes over 50,000 cases <strong>of</strong> bacteremia <strong>and</strong> at least 500,000 cases <strong>of</strong> pneumonia annually in<br />
the United States. 1-3 Although pneumococcus is an important pathogen in meningitis, bronchitis,<br />
<strong>and</strong> otitis media, these disease processes will not be discussed in this report.<br />
The epidemiologic rationale for targeting pneumococcal vaccination among hospitalized<br />
patients derives from research showing that two-thirds <strong>of</strong> patients hospitalized with serious<br />
pneumococcal infections had been hospitalized at least once in the previous 3-5 years. 4-6 One<br />
retrospective study showed that about 60% <strong>of</strong> persons 65 years <strong>of</strong> age <strong>and</strong> older hospitalized<br />
with pneumonia had been discharged from a hospital at least once in the prior 4 years. 4 A<br />
prospective cohort study <strong>of</strong> patients aged ≥65 discharged from the hospital showed that the 5year<br />
probability for readmission with pneumonia was over 7%. 4 Hospitalization, therefore, is a<br />
marker for patients at increased risk <strong>of</strong> developing subsequent pneumococcal infection. Despite<br />
the scope <strong>of</strong> the problem <strong>and</strong> the appeals for action from multiple specialty societies <strong>and</strong> national<br />
health care organizations, the vaccine is underutilized in the inpatient setting. 5-7 This<br />
underutilization has been attributed to uncertain effectiveness <strong>of</strong> the vaccine <strong>and</strong> ineffective<br />
methods <strong>of</strong> vaccine delivery during hospitalization.<br />
Hospital-based vaccination for patients at high risk <strong>of</strong> contracting pneumococcal<br />
infections is part <strong>of</strong> the action plan for adult immunizations developed by the Federal Centers for<br />
Disease Control <strong>and</strong> Prevention (CDC) <strong>and</strong> the Health Care Financing Administration (HCFA). 6<br />
It is also endorsed by the CDC’s Advisory Committee on Immunization <strong>Practices</strong> (ACIP) 5, 6 <strong>and</strong><br />
the National Vaccine Advisory Committee. In addition, most national guidelines for the<br />
management <strong>of</strong> patients hospitalized with community-acquired pneumonia recommend<br />
vaccination against pneumococcus at time <strong>of</strong> discharge. 1,8<br />
Practice Description<br />
Currently, pneumococcal vaccines contain 23 capsular polysaccharide antigens <strong>of</strong> S.<br />
pneumoniae (23-valent vaccines). Over 88% <strong>of</strong> the serotypes that cause invasive disease in the<br />
United States, as well as 88% <strong>of</strong> serotypes accounting for penicillin-resistant isolates, are<br />
included in the 23-valent vaccine. 2,9 Newer conjugate vaccines designed primarily to enhance the<br />
immune response in children are not covered in this review.<br />
The practice <strong>of</strong> hospital-based pneumococcal vaccination is recommended for patients at<br />
increased risk for pneumococcal infection or increased risk <strong>of</strong> experiencing severe disease.<br />
<strong>Patient</strong>s ≥65 years <strong>of</strong> age, or patients with certain chronic illnesses, including chronic<br />
cardiovascular disease, chronic pulmonary disease, diabetes, alcoholism, chronic liver disease,<br />
<strong>and</strong> functional or anatomic asplenia, are deemed high-risk. 3 It is also recommended that all<br />
immunocompromised patients (due to HIV infection, leukemia, lymphoma, long term steroid<br />
use, or organ transplantation, among other causes) <strong>and</strong> any patient admitted with a diagnosis <strong>of</strong><br />
community-acquired pneumonia be vaccinated. 1,3,8 Vaccination could occur at any time during<br />
the hospitalization, but is <strong>of</strong>ten recommended at discharge.<br />
385
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The goal <strong>of</strong> pneumococcal vaccination in hospitalized patients is to reduce the morbidity<br />
<strong>and</strong> mortality associated with pneumococcal infection, namely pneumococcal bacteremia <strong>and</strong><br />
pneumococcal pneumonia. The CDC estimates the annual incidence <strong>of</strong> pneumococcal<br />
bacteremia at 15-30 cases per 100,000 population <strong>and</strong> 50-83 cases per 100,000 in persons aged<br />
≥65. 3 A recent study <strong>of</strong> the epidemiology <strong>of</strong> invasive S. pneumoniae (ie, associated with<br />
bacteremia) in the United States found an overall incidence <strong>of</strong> 23.2 cases per 100,000,<br />
corresponding to 62,840 cases annually. 2 The incidence among adults aged 65 <strong>and</strong> older was<br />
59.7 per 100,000. The overall fatality rate was 10%, but patients aged 18-64 with an ACIP<br />
indication for vaccination had a fatality rate <strong>of</strong> 12.1%. <strong>Patient</strong>s ≥65 years <strong>of</strong> age accounted for<br />
51.4% <strong>of</strong> all deaths. These figures result in national estimates <strong>of</strong> over 6000 deaths in 1998<br />
attributed to invasive pneumococcal disease. 2<br />
The precise incidence <strong>of</strong> pneumococcal pneumonia is harder to estimate due to the poor<br />
sensitivity <strong>and</strong> specificity <strong>of</strong> diagnostic tests for this disease. At least 25-35% <strong>of</strong> all pneumonias<br />
are linked to S. pneumoniae, resulting in a minimum <strong>of</strong> 500,000 cases <strong>of</strong> pneumococcal<br />
pneumonia annually. Bacteremia complicates pneumococcal pneumonia in 10-25% <strong>of</strong> cases. 3<br />
The mortality rate for all patients hospitalized with community-acquired pneumonia is estimated<br />
at between 10-15%. 1<br />
Opportunities for Impact<br />
Despite recommendations to routinely vaccinate eligible hospitalized patients,<br />
pneumococcal vaccine is underutilized. The high potential impact <strong>of</strong> vaccination is borne out by<br />
recent epidemiologic evidence. Based on 1998 projections, 76% <strong>of</strong> invasive pneumococcal<br />
disease <strong>and</strong> 87% <strong>of</strong> deaths occur in patients who are eligible for pneumococcal vaccine. 2 In<br />
addition, 88% <strong>of</strong> penicillin-resistant isolates during the same time period were <strong>of</strong> serotypes<br />
included in the 23-valent vaccine. 9<br />
The vaccine is currently recommended for over 30 million persons aged ≥ 65 <strong>and</strong> over 23<br />
million persons
<strong>and</strong>omized controlled trials (RCTs) <strong>of</strong> vaccines (valences ranging from 6 to 17) in adults with<br />
<strong>and</strong> without risk factors for pneumococcal infection. Results were pooled <strong>and</strong> analyzed for<br />
effects in various subgroups with careful attention to study heterogeneity (rate differences (RD)<br />
were reported when significant heterogeneity existed). 10<br />
The second study 11 included 13 r<strong>and</strong>omized <strong>and</strong> quasi-r<strong>and</strong>omized studies <strong>of</strong> vaccines<br />
with valences ≥2. Consequently, this study 11 included 2 quasi-r<strong>and</strong>omized studies with vaccine<br />
valences
overall or in high-risk patients (ie, patients age 55 years, patients with one or more chronic<br />
medical problems, <strong>and</strong> immunocompromised patients). Results for pneumococcal infectionrelated<br />
outcomes did achieve significance in low-risk patients. 10<br />
Hutchison et al found effectiveness against pneumococcal pneumonia ranging from 31%<br />
to 76% (results were significant in 3 trials), but study heterogeneity prevented a pooled<br />
estimate. 11 Regression analysis suggested similar benefits for systemic pneumococcal infection<br />
in elderly patients, but were inconclusive for systemic infection in chronically ill patients. For<br />
elderly patients, the authors estimated a number needed to treat (NNT) <strong>of</strong> 2520 to prevent a<br />
single case <strong>of</strong> pneumococcal bacteremia per year. 11<br />
The most recent meta-analysis 12 reported that in 3 comparisons involving approximately<br />
21,100 immunocompetent subjects, pneumococcal vaccination was associated with significant<br />
reductions in the incidence <strong>of</strong> all-cause pneumonia (relative risk 0.56, 95% CI: 0.47-0.66),<br />
pneumococcal pneumonia (relative risk 0.16, 95% CI: 0.11-0.23), pneumonia deaths (relative<br />
risk 0.70, 95% CI: 0.50-0.96) <strong>and</strong> bacteremia (relative risk 0.18, 95% CI: 0.09-0.34). However,<br />
in 10 comparisons involving over 24,000 subjects who were elderly or likely to have impaired<br />
immune systems, the authors found no benefit to pneumococcal vaccination in terms <strong>of</strong> any<br />
clinical outcome <strong>of</strong> interest. 12 While the relative risk for pneumococcal bacteremia in elderly or<br />
high-risk patients showed a trend towards benefit, the results were not statistically significant<br />
(relative risk 0.53, 95% CI: 0.14-1.94).<br />
One additional publication 21 has appeared since the search period covered by this most<br />
recent meta-analysis. 12 This publication represents a 6-month preliminary report from a<br />
prospective comparison between 2 large cohorts (>100,000 subjects each) <strong>of</strong> Swedish patients<br />
age 65 years. <strong>Patient</strong>s in one group received pneumococcal vaccine, influenza vaccine, or both.<br />
The other cohort consisted <strong>of</strong> all patients from the same region <strong>and</strong> age group who chose not to<br />
receive either <strong>of</strong> the vaccines. Among all vaccinated patients (results are pooled for<br />
pneumococcal <strong>and</strong> influenza vaccines), hospital admission for pneumonia (including all-cause<br />
pneumonia, pneumococcal pneumonia <strong>and</strong> invasive pneumococcal pneumonia) was significantly<br />
reduced, <strong>and</strong> overall mortality was reduced by 57% (95% CI: 55-60%). This study design has<br />
significant potential for bias, in that people who elect to participate in clinical studies tend to<br />
have better outcomes independent <strong>of</strong> the treatments they receive. 22-25 Nonetheless, it seems<br />
unlikely that the results observed in this study, including a 57% reduction in all-cause mortality<br />
over a 6-month period, could be attributable solely to a selection effect tied to patients’ decision<br />
to participate in the study.<br />
36.2. Vaccine Delivery Methods<br />
Study Designs <strong>and</strong> Outcomes<br />
Multiple studies have evaluated the effectiveness <strong>of</strong> various methods <strong>of</strong> increasing rates<br />
<strong>of</strong> vaccination among eligible patients in both the inpatient <strong>and</strong> outpatient settings. We focused<br />
our review on interventions relating to inpatient settings (ie, hospitals <strong>and</strong> nursing homes). Two<br />
systematic reviews published in 1994 <strong>and</strong> 1999 evaluate multiple strategies. 26, 27 The first review<br />
identified 3 studies in hospitalized patients <strong>and</strong> one in institutionalized patients. 26 The second<br />
review includes studies identified in the first, <strong>and</strong> comments on several additional studies. 27 Both<br />
reviews grade the included studies <strong>and</strong> report pooled estimates <strong>of</strong> effectiveness. The 2 reviews<br />
use slightly different definitions <strong>of</strong> the types <strong>of</strong> interventions, but are internally consistent with<br />
respect to the most effective strategy, namely st<strong>and</strong>ing orders.<br />
388
All studies <strong>of</strong> the effectiveness <strong>of</strong> vaccine delivery methods reported vaccination rates.<br />
The results <strong>of</strong> several heterogeneous studies were pooled. There are few RCTs included in the<br />
summary estimates, <strong>and</strong> most interventions were studied with a before-after analysis <strong>of</strong><br />
vaccination rates. Where possible, we report pooled estimates for given methods <strong>of</strong> improving<br />
vaccine delivery, <strong>and</strong> then comment on individual studies with the most promise for improving<br />
vaccination rates. No study <strong>of</strong> delivery methods looked at clinical outcomes.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The systematic reviews <strong>of</strong> vaccine delivery methods with provider reminders in the<br />
inpatient setting were associated with absolute increases in vaccination rates that ranged from<br />
7.5%-17% (Table 36.2.1). 26,27 One subsequent before-after study <strong>of</strong> chart reminders in<br />
hospitalized patients showed that vaccination rates in eligible patients increased from 0% to<br />
28.8%. 28 The most impressive effects were seen for system-related changes, which increased<br />
vaccination rates by 45-51%. The most effective system-related change was the implementation<br />
<strong>of</strong> st<strong>and</strong>ing orders, which produced increases <strong>of</strong> 69-81% over usual care. 27 Studies <strong>of</strong> pneumonia<br />
clinical pathways (Chapter 52) have shown no effect on pneumococcal vaccination rates despite<br />
improvements in other pathway processes. 7<br />
Potential for Harm (from pneumococcal vaccination)<br />
Three <strong>of</strong> the studies analyzed by Fine et al 10 reported data on adverse effects. Erythema<br />
ranged from 30.6-35.1% in the vaccine group compared with 1.7-3.5% in the control group.<br />
Fever developed in 2.0% <strong>of</strong> vaccinated patients compared with 1.2% <strong>of</strong> controls. No fatal or lifethreatening<br />
adverse events occurred. Moore et al 12 report that in one study, in addition to<br />
increased rates <strong>of</strong> fever, vaccine recipients were more likely to experience a swollen or sore arm.<br />
There is the theoretical concern that patients vaccinated in the hospital may have an increased<br />
chance <strong>of</strong> being inappropriately revaccinated if they are unaware <strong>of</strong> their prior vaccination status<br />
(due to illness, etc.) or are being cared for by a physician other than their primary care doctor. A<br />
recent study compared the safety <strong>of</strong> the vaccine in patients receiving a first vaccination <strong>and</strong><br />
patients receiving re-vaccination 5 years after their prior dose. There was an increase in selflimited<br />
local reactions with re-vaccination (RR 3.3, 95% CI: 2.1-5.1), but no serious adverse<br />
reactions were reported. 29 Few data address rates <strong>of</strong> adverse reactions in hospitalized patients<br />
vaccinated at discharge who receive re-vaccination earlier than 5 years after their prior dose.<br />
Finally, the recent study in Ug<strong>and</strong>an HIV-infected adults showed trends toward increased<br />
rates <strong>of</strong> invasive pneumococcal disease, all pneumococcal events, <strong>and</strong> a statistically significant<br />
increase in all-cause pneumonia (hazard ratio 1.89, 95% CI: 1.1-3.2) among vaccine recipients. 19<br />
This study calls into question the utility <strong>of</strong> giving pneumococcal vaccine to HIV-infected<br />
individuals.<br />
Costs <strong>and</strong> Implementation<br />
The cost-effectiveness <strong>of</strong> hospital-based pneumococcal vaccination is difficult to<br />
determine in light <strong>of</strong> the debate over how effective the vaccine is in reducing pneumococcal<br />
outcomes in at-risk populations. A cost-effectiveness analysis <strong>of</strong> vaccination for all elderly<br />
patients in the United States recently demonstrated a wide range <strong>of</strong> possible outcomes that<br />
depended on assumptions <strong>of</strong> vaccine effectiveness <strong>and</strong> duration <strong>of</strong> protection. 30 Base-case<br />
estimates showed that vaccination was cost saving at $8.27, <strong>and</strong> gained 1.21 quality-adjusted<br />
days <strong>of</strong> life per person vaccinated. Factoring in the future medical costs <strong>of</strong> survivors, vaccinating<br />
all patients ≥ 65 years <strong>of</strong> age would cost $9600 per quality-adjusted life year under the most<br />
389
optimistic assumptions about vaccine effectiveness, <strong>and</strong> $51,661 per quality-adjusted life-year<br />
under worst-case assumptions. 30 A recent systematic review <strong>of</strong> pneumococcal vaccine costeffectiveness<br />
by authors from the Cochrane Vaccines Field concluded that there is too much<br />
variability in assessments <strong>of</strong> cost-effectiveness (largely attributed to uncertainty over vaccine<br />
effectiveness) to reach any firm conclusions. 31 The authors called for a moratorium on all<br />
economic modeling until completion <strong>of</strong> a Cochrane review <strong>of</strong> pneumococcal vaccine<br />
effectiveness.<br />
Comment<br />
Pneumococcal vaccine is effective in reducing invasive disease in low-risk patients. It<br />
appears effective in reducing invasive disease in the elderly <strong>and</strong> high-risk patients based on<br />
results <strong>of</strong> one meta-analysis, multiple case-control studies, <strong>and</strong> a CDC serotype prevalence<br />
study. Importantly however, 2 meta-analyses failed to demonstrate a significant benefit <strong>of</strong> the<br />
vaccine for any outcomes in elderly or other high-risk patients. The vaccine appears efficacious<br />
in non-bacteremic disease (pneumococcal pneumonia) in low-risk patients, <strong>and</strong> one metaanalysis<br />
suggests a protective effect against pneumococcal pneumonia in the elderly, 11 but the<br />
others do not. 10,12 No study has demonstrated reductions in mortality. Thus, in the population<br />
most likely to be targeted by hospital-based immunization programs (elderly, high-risk) vaccine<br />
efficacy remains inconclusive. If it is assumed to be effective in reducing the incidence <strong>of</strong> the<br />
most serious outcome in this population (pneumococcal bacteremia), best estimates suggest a<br />
very large NNT (>2500).<br />
Increasing antibiotic resistance, the aging <strong>of</strong> the US population <strong>and</strong> the major burden <strong>of</strong><br />
pneumococcal disease among adults <strong>and</strong> elderly patients make increasing vaccination rates an<br />
obvious goal if, in fact, the vaccine is effective. The evidence supports system changes (in<br />
particular, the use <strong>of</strong> st<strong>and</strong>ing orders) as the best method <strong>of</strong> increasing vaccine rates in eligible,<br />
hospitalized patients. Though such rates can be increased, available data regarding vaccine<br />
efficacy raise doubts about the overall utility <strong>of</strong> both local <strong>and</strong> national initiatives aimed at<br />
increasing the rates <strong>of</strong> pneumococcal vaccination at hospital discharge. Early enthusiasm<br />
explains the large number <strong>of</strong> national pneumococcal vaccine initiatives, however available data<br />
on effectiveness provide only modest support for these initiatives.<br />
390
Table 36.1. 1. Pneumococcal vaccine efficacy*<br />
Study Description Study Design,<br />
Outcomes<br />
Meta-analysis <strong>of</strong> 9 RCTs.<br />
Vaccine valences 6-17,<br />
included international<br />
studies, high- <strong>and</strong> lowrisk<br />
patients. 10<br />
Meta-analysis <strong>of</strong> 13 RCTs<br />
<strong>and</strong> “quasi-r<strong>and</strong>omized”<br />
trials. Vaccine valences 2-<br />
17, includes international<br />
trials, high <strong>and</strong> low-risk<br />
patients. 11<br />
Meta-analysis <strong>of</strong> 13 RCTs<br />
including three recent<br />
trials published after the<br />
last meta-analysis. 12<br />
Six-month preliminary<br />
results report from<br />
prospective comparison<br />
between cohort <strong>of</strong><br />
100,242 Swedish patients<br />
age 65 years who<br />
received pneumovax,<br />
influenza vaccine, or<br />
both, <strong>and</strong> patients from<br />
the same region <strong>and</strong> age<br />
group who chose not to<br />
participate in the study. 21<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 1A,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Results (95% Confidence Interval)<br />
* OR indicates odds ratio; RD, risk difference; <strong>and</strong> RR, relative risk.<br />
Definitive pneumococcal pneumonia:<br />
OR 0.34 (0.24-0.48), RD 4/1000 (0-7)<br />
Definitive pneumococcal pneumonia, vaccine types:<br />
OR 0.17 (0.09-0.33), RD 8/1000 (1-16)<br />
Presumptive pneumococcal pneumonia:<br />
OR 0.47 (0.35-0.63), RD 13/1000 (-21 to 47)<br />
All cause pneumonia: OR 0.90 (0.77-1.04)<br />
Mortality: OR 1.02 (0.90-1.14)<br />
Stratified results show no benefit in any outcome for highrisk<br />
patients.<br />
Systemic infection, vaccine type: OR 0.17 (0.09-0.31)<br />
Systemic infection all types: OR 0.27 (0.13-0.49)<br />
All cause pneumococcal pneumonia: OR range 0.24-0.69,<br />
results significant in 3 studies<br />
Pneumococcal pneumonia, vaccine types:<br />
OR range 0.08-0.85, 8 <strong>of</strong> 9 studies showed reduced risk,<br />
results significant in 6 studies.<br />
Stratified results show benefit in elderly, mixed results in<br />
chronically ill.<br />
All pneumonias:<br />
Healthy (H), RR 0.56 (0.47-0.66), NNT 29 (24-36);<br />
Elderly, High Risk (E) RR 1.08 (0.92-1.27)<br />
Pneumococcal pneumonias:<br />
H, RR 0.16 (0.11-0.23), NNT 38 (33-45);<br />
E, RR 0.88 (0.72-1.07)<br />
Pneumococcal bacteremia:<br />
H, RR 0.18 (0.09-0.34), NNT 32 (26-44);<br />
E, (3 trials <strong>and</strong> only 927 pts), RR 0.53 (0.14-1.94)<br />
Pneumonia-related death:<br />
H, RR 0.70 (0.50-0.96), NNT 213 (114-1660);<br />
E, RR 0.93 (0.72-1.20)<br />
Results pool all vaccinated patients.<br />
Hospital admission for all cause pneumonia reduced by 29%<br />
(24-34), for pneumococcal pneumonia 36% (3-58),<br />
invasive pneumococcal pneumonia 52% (1-77).<br />
Overall mortaliy reduced by 57% (55-60)<br />
391
Table 36.2. 1. Vaccine delivery<br />
Study Description Study Design,<br />
Outcomes<br />
Systematic review <strong>of</strong> studies Level 2A-3A<br />
from 1980-1997 <strong>of</strong> methods to Vaccination rates<br />
increase vaccination rates. (Level 2)<br />
Multiple vaccines (eg,<br />
pneumococcal, influenza,<br />
hepatitis) delivered in<br />
multiple inpatient <strong>and</strong><br />
ambulatory settings were<br />
reviewed, but this summary<br />
focuses on pneumococcal<br />
vaccine in hospitalized<br />
patients<br />
Systematic review <strong>of</strong> studies Level 2A-3A<br />
from 1979-1992. Multiple Vaccination rates<br />
vaccine types in multiple (Level 2)<br />
settings.<br />
References<br />
392<br />
Results<br />
Provider reminders increased vaccination rates<br />
by 17% (pooled absolute increases for all<br />
vaccines in all settings, with a range <strong>of</strong> 1-<br />
67%).<br />
St<strong>and</strong>ing orders achieved a 51% mean absolute<br />
increase in vaccination rates (range 30-81%)<br />
for all vaccine types; pneumococcal<br />
vaccination in particular the increases ranged<br />
from 69% to 81% (in hospital setting <strong>and</strong> longterm<br />
care facility).<br />
Provider-oriented interventions resulted in a<br />
7.5% increase in vaccination coverage (3.4-<br />
11.6%)<br />
System-oriented interventions resulted in a<br />
45.5% increase in vaccination coverage (95%<br />
CI: 37.2-53.7%).<br />
1. Bartlett JG, Dowell SF, M<strong>and</strong>ell LA, File Jr TM, Musher DM, Fine MJ. Practice guidelines<br />
for the management <strong>of</strong> community-acquired pneumonia in adults. Infectious Diseases<br />
Society <strong>of</strong> America. Clin Infect Dis. 2000;31:347-382.<br />
2. Robinson KA, Baughman W, Rothrock G, Barrett NL, Pass M, Lexau C, et al.<br />
Epidemiology <strong>of</strong> invasive Streptococcus pneumoniae infections in the United States, 1995-<br />
1998: Opportunities for prevention in the conjugate vaccine era. JAMA. 2001;285:1729-<br />
1735.<br />
3. Prevention <strong>of</strong> pneumococcal disease: recommendations <strong>of</strong> the Advisory Committee on<br />
Immunization <strong>Practices</strong> (ACIP). MMWR Morb Mortal Wkly Rep. 1997;46:1-24.<br />
4. Fedson DS, Harward MP, Reid RA, Kaiser DL. Hospital-based pneumococcal<br />
immunization. Epidemiologic rationale from the Shen<strong>and</strong>oah study. JAMA.<br />
1990;264:1117-1122.<br />
5. Missed opportunities for pneumococcal <strong>and</strong> influenza vaccination <strong>of</strong> Medicare pneumonia<br />
inpatients–12 western states, 1995. MMWR Morb Mortal Wkly Rep. 1997;46:919-923.<br />
6. Fedson DS, Houck P, Bratzler D. Hospital-based influenza <strong>and</strong> pneumococcal vaccination:<br />
Sutton's Law applied to prevention. Infect Control Hosp Epidemiol. 2000;21:692-699.<br />
7. Metersky ML, Fine JM, Tu GS, Mathur D, Weingarten S, Petrillo MK, et al. Lack <strong>of</strong> effect<br />
<strong>of</strong> a pneumonia clinical pathway on hospital-based pneumococcal vaccination rates. Am J<br />
Med. 2001;110:141-143.<br />
8. Heffelfinger JD, Dowell SF, Jorgensen JH, Klugman KP, Mabry LR, Musher DM, et al.<br />
Management <strong>of</strong> community-acquired pneumonia in the era <strong>of</strong> pneumococcal resistance: a<br />
report from the Drug-Resistant Streptococcus pneumoniae Therapeutic Working Group.<br />
Arch Intern Med. 2000;160:1399-1408.
9. Whitney CG, Farley MM, Hadler J, Harrison LH, Lexau C, Reingold A, et al. Increasing<br />
prevalence <strong>of</strong> multidrug-resistant Streptococcus pneumoniae in the United States. N Engl J<br />
Med. 2000;343:1917-1924.<br />
10. Fine MJ, Smith MA, Carson CA, Meffe F, Sankey SS, Weissfeld LA, et al. Efficacy <strong>of</strong><br />
pneumococcal vaccination in adults. A meta-analysis <strong>of</strong> r<strong>and</strong>omized controlled trials. Arch<br />
Intern Med. 1994;154:2666-2677.<br />
11. Hutchison BG, Oxman AD, Shannon HS, Lloyd S, Altmayer CA, Thomas K. Clinical<br />
effectiveness <strong>of</strong> pneumococcal vaccine. Meta-analysis. Can Fam Physician. 1999;45:2381-<br />
2393.<br />
12. Moore RA, Wiffen PJ, Lipsky BA. Are the pneumococcal polysaccharide vaccines<br />
effective? Meta-analysis <strong>of</strong> the prospective trials. BMC Fam Pract. 2000;1:1.<br />
13. Austrian R. Surveillance <strong>of</strong> pneumoccal infection for field trials <strong>of</strong> polyvalent<br />
pneumococcal vaccines. Bethesda MD. 1980: National Institute <strong>of</strong> Health; 1980. NIH<br />
<strong>Public</strong>ation DAB-VDP-12-84. Contract No 1A13257. 1-59.<br />
14. MacLoed CM, Hodges RG, Heidelberger M, Bernhard WG. CM MacLoed, RG Hodges, M<br />
Heidelberger, WG Bernhard: Prevention <strong>of</strong> pneumococcal pneumonia by immunization<br />
with specific capsular polysaccharides. J Exp Med. 1945;82:445-465.<br />
15. Kaufman P. Pneumonia in old age: active immunization against pneumonia with<br />
pneumococcus polysaccharide, results <strong>of</strong> a six year study. Arch Intern Med. 1947;79:518-<br />
531.<br />
16. Honkanen PO, Keistinen T, Miettinen L, Herva E, Sankilampi U, Laara E, et al.<br />
Incremental effectiveness <strong>of</strong> pneumococcal vaccine on simultaneously administered<br />
influenza vaccine in preventing pneumonia <strong>and</strong> pneumococcal pneumonia among persons<br />
aged 65 years or older. Vaccine. 1999;17:2493-2500.<br />
17. Koivula I, Sten M, Leinonen M, Makela PH. Clinical efficacy <strong>of</strong> pneumococcal vaccine in<br />
the elderly: a r<strong>and</strong>omized, single-blind population-based trial. Am J Med. 1997;103:281-<br />
290.<br />
18. Ortqvist A, Hedlund J, Burman LA, Elbel E, H<strong>of</strong>er M, Leinonen M, et al. R<strong>and</strong>omised trial<br />
<strong>of</strong> 23-valent pneumococcal capsular polysaccharide vaccine in prevention <strong>of</strong> pneumonia in<br />
middle-aged <strong>and</strong> elderly people. Swedish Pneumococcal Vaccination Study Group. Lancet.<br />
1998;351:399-403.<br />
19. French N, Nakiyingi J, Carpenter LM, Lugada E, Watera C, Moi K, et al. 23-valent<br />
pneumococcal polysaccharide vaccine in HIV-1-infected Ug<strong>and</strong>an adults: double-blind,<br />
r<strong>and</strong>omised <strong>and</strong> placebo controlled trial. Lancet. 2000;355:2106-2111.<br />
20. Butler JC, Breiman RF, Campbell JF, Lipman HB, Broome CV, Facklam RR.<br />
Pneumococcal polysaccharide vaccine efficacy. An evaluation <strong>of</strong> current<br />
recommendations. JAMA. 1993;270:1826-1831.<br />
21. Christenson B, Lundbergh P, Hedlund J, Ortqvist A. Effects <strong>of</strong> a large-scale intervention<br />
with influenza <strong>and</strong> 23-valent pneumococcal vaccines in adults aged 65 years or older: a<br />
prospective study. Lancet. 2001;357:1008-1011.<br />
22. Davis S, Wright PW, Schulman SF, Hill LD, Pinkham RD, Johnson LP, et al. Participants<br />
in prospective, r<strong>and</strong>omized clinical trials for resected non-small cell lung cancer have<br />
improved survival compared with nonparticipants in such trials. Cancer. 1985;56:1710-<br />
1718.<br />
23. Pirie PL, Elias WS, Wackman DB, Jacobs DR, Jr., Murray DM, Mittelmark MB, et al.<br />
Characteristics <strong>of</strong> participants <strong>and</strong> nonparticipants in a community cardiovascular disease<br />
risk factor screening: the Minnesota Heart Health Program. Am J Prev Med. 1986;2:20-25.<br />
393
24. Jha P, Deboer D, Sykora K, Naylor CD. Characteristics <strong>and</strong> mortality outcomes <strong>of</strong><br />
thrombolysis trial participants <strong>and</strong> nonparticipants: a population-based comparison. J Am<br />
Coll Cardiol. 1996;27:1335-1342.<br />
25. Kessenich CR, Guyatt GH, Rosen CJ. Health-related quality <strong>of</strong> life <strong>and</strong> participation in<br />
osteoporosis clinical trials. Calcif Tissue Int. 1998;62:189-192.<br />
26. Gyorkos TW, Tannenbaum TN, Abrahamowicz M, Bedard L, Carsley J, Franco ED, et al.<br />
Evaluation <strong>of</strong> the effectiveness <strong>of</strong> immunization delivery methods. Can J <strong>Public</strong> Health.<br />
1994;85(Suppl 1):S14-30.<br />
27. Shefer A, Briss P, Rodewald L, Bernier R, Strikas R, Yusuf H, et al. Improving<br />
immunization coverage rates: an evidence-based review <strong>of</strong> the literature. Epidemiol Rev.<br />
1999;21:96-142.<br />
28. Vondracek TG, Pham TP, Huycke MM. A hospital-based pharmacy intervention program<br />
for pneumococcal vaccination. Arch Intern Med. 1998;158:1543-1547.<br />
29. Jackson LA, Benson P, Sneller VP, Butler JC, Thompson RS, Chen RT, et al. <strong>Safety</strong> <strong>of</strong><br />
revaccination with pneumococcal polysaccharide vaccine. JAMA. 1999;281:243-248.<br />
30. Sisk JE, Moskowitz AJ, Whang W, Lin JD, Fedson DS, McBean AM, et al. Costeffectiveness<br />
<strong>of</strong> vaccination against pneumococcal bacteremia among elderly people.<br />
JAMA. 1997;278:1333-1339.<br />
31. Hutton J, Iglesias C, Jefferson TO. Assessing the potential cost effectiveness <strong>of</strong><br />
pneumococcal vaccines in the US: methodological issues <strong>and</strong> current evidence. Drugs<br />
Aging. 1999;15:31-36.<br />
394
395
Chapter 37. Pain Management<br />
Erica Brownfield, MD<br />
Emory University School <strong>of</strong> Medicine<br />
Approximately 23 million people undergo surgery each year in the United States. 1<br />
Despite pharmacologic interventions, at least 40-50% <strong>of</strong> postoperative patients report inadequate<br />
pain relief. 2 In addition, the practice <strong>of</strong> withholding analgesics due to fear <strong>of</strong> masking<br />
symptomatology <strong>and</strong> delaying diagnosis is still widespread in many emergency rooms <strong>and</strong> acute<br />
care settings. The Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations (JCAHO) <strong>and</strong><br />
the Agency for Health Care Policy <strong>and</strong> Research (AHCPR) have established guidelines for the<br />
appropriate assessment <strong>and</strong> management <strong>of</strong> pain in general <strong>and</strong> postoperatively. Yet efforts to<br />
educate clinicians as to appropriate pain management, particularly in emergency departments<br />
(ED) <strong>and</strong> following surgery, have lagged behind the available evidence.<br />
We have taken the point <strong>of</strong> view that untreated pain represents a patient safety problem.<br />
This chapter reviews pain management techniques <strong>and</strong> interventions in 4 domains: use <strong>of</strong><br />
analgesics in patients with acute abdominal pain, the use <strong>of</strong> acute pain services, prophylactic<br />
antiemetics during patient-controlled analgesia therapy, <strong>and</strong> non-pharmacologic interventions for<br />
postoperative pain.<br />
Subchapter 37.1. Use <strong>of</strong> Analgesics in the Acute Abdomen<br />
Background<br />
The use <strong>of</strong> analgesics in patients with acute abdominal pain has traditionally been<br />
condemned. The 1987 edition <strong>of</strong> Cope’s Early Diagnosis <strong>of</strong> the Acute Abdomen states “though it<br />
may appear crude, it is really prudent to withhold morphine until a reasonable diagnosis has been<br />
made <strong>and</strong> a plan <strong>of</strong> action formulated.” 1 The most recent edition <strong>of</strong> Cope’s Early Diagnosis <strong>of</strong><br />
the Acute Abdomen (1996) begins to question this long-accepted dogma, but still states that<br />
analgesia medication should be given only after a “responsible surgeon” takes a thorough history<br />
<strong>and</strong> performs a thorough physical examination. 3 As patients with acute abdominal pain are rarely<br />
evaluated by a surgeon within the first few hours <strong>of</strong> presentation, it seems inappropriate <strong>and</strong><br />
inhumane to withhold pain medication if this practice is not supported by evidence.<br />
Practice Description<br />
Prescribing analgesics to patients with acute abdominal pain is infrequently done. When<br />
prescribed, dosages <strong>and</strong> routes vary, from intramuscular to intravenous morphine in 5 to 10 mg<br />
increments, or 0.1 mg/kg <strong>of</strong> body weight. Although JCAHO <strong>and</strong> AHCPR have established<br />
guidelines for the appropriate assessment <strong>and</strong> management <strong>of</strong> pain in general <strong>and</strong><br />
postoperatively, neither addresses pain management in patients with acute abdomens. Therefore,<br />
the traditional practice <strong>of</strong> withholding analgesia in this setting has not been seriously challenged.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
According to the National Center <strong>of</strong> Health Statistics, there were 100,385,000 total visits<br />
to US emergency departments in 1998. The most frequent principal reason for visits was<br />
stomach or abdominal pain (5.9 million). 4 Despite studies suggesting that the early<br />
administration <strong>of</strong> pain medication is safe <strong>and</strong> does not interfere with, <strong>and</strong> may actually facilitate,<br />
the ability to make a correct diagnosis, recent surveys <strong>of</strong> emergency room physicians <strong>and</strong><br />
396
surgeons indicate that the majority withhold analgesics in patients presenting with an acute<br />
abdomen. Wolfe et al 2 surveyed 443 emergency medicine physicians <strong>and</strong> found that although<br />
85% believe that the conservative administration <strong>of</strong> pain medication did not change important<br />
physical findings, 76% choose not to give an opiate analgesic until after the examination by a<br />
surgeon. Graber et al 5 surveyed 131 practicing surgeons in Iowa <strong>and</strong> found 67% agreed that pain<br />
medications interfere with diagnostic accuracy, <strong>and</strong> 82% cited their concerns about diagnostic<br />
accuracy when deciding to withhold pain medication.<br />
Opportunities for Impact<br />
Limited data suggest that the number <strong>of</strong> patients with acute abdominal pain who actually<br />
receive pain relief before surgical evaluation is small. Therefore, by educating providers in<br />
appropriate pain management for these patients, the potential impact in emergency departments<br />
across the country is large.<br />
Study Designs<br />
Five prospective r<strong>and</strong>omized controlled trials (Level 1) were evaluated. Four <strong>of</strong> the 5<br />
used a double-blind design. In each study, patients were r<strong>and</strong>omly assigned to receive opiate<br />
analgesia or placebo, <strong>and</strong> evaluated pre- <strong>and</strong> post-intervention for pain using variations <strong>of</strong> visual<br />
analog scales (Table 37.1.1).<br />
Study Outcomes<br />
All 5 studies evaluated the effects <strong>of</strong> analgesia on pain relief (Level 1), <strong>and</strong> diagnoses <strong>and</strong><br />
treatment decisions (Level 2) in patients with acute abdominal pain. Two studies evaluated the<br />
effects <strong>of</strong> analgesia on physical examination findings <strong>and</strong> one evaluated the effects <strong>of</strong> analgesia<br />
on the diagnostic performance <strong>of</strong> ultrasonography in patients with acute abdominal pain (Level<br />
3).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
All 5 studies showed that provision <strong>of</strong> analgesia decreased pain more than it decreased<br />
localization <strong>of</strong> tenderness. None <strong>of</strong> the 5 studies indicate that the practice <strong>of</strong> providing early<br />
analgesia is harmful. Specifically, no study found compromises in diagnosis or treatment <strong>of</strong> the<br />
acute abdomen after increasing the use <strong>of</strong> analgesia.<br />
Potential for Harm<br />
The traditional belief that analgesic use in patients with acute abdominal pain may mask<br />
signs <strong>and</strong> symptoms, delay diagnosis, <strong>and</strong> lead to increased morbidity <strong>and</strong> mortality was not<br />
supported in these studies. All 5 studies analyzed diagnostic or management errors that occurred<br />
in each group.<br />
Attard et al 6 found no difference in localization <strong>of</strong> physical signs, <strong>and</strong> no difference in the<br />
surgeon’s diagnostic confidence or management decision (to operate or to observe) between the<br />
2 groups (opioids vs. placebo). The decision to operate or to observe was incorrect in 2 patients<br />
in the opioid group (4%) <strong>and</strong> in 9 patients in the placebo group (18%). The surgeon’s initial<br />
diagnosis one hour after the injection was incorrect in all <strong>of</strong> these patients. These same 2 patients<br />
in the opioid group were incorrectly diagnosed as having non-specific abdominal pain when first<br />
assessed, but the diagnosis was subsequently changed <strong>and</strong> both patients had an inflamed<br />
appendix removed within 24 hours <strong>of</strong> admission. Neither appendix was perforated. There were<br />
no deaths or side effects from the injection for either group.<br />
397
LoVecchio et al 7 documented changes in localization <strong>of</strong> physical examination findings<br />
<strong>and</strong> differences in diagnosis between patients receiving opioid analgesia <strong>and</strong> placebo. The use <strong>of</strong><br />
opioids was associated with some change in tenderness <strong>and</strong> localization in half the patients but<br />
led to no delays in care or eventual morbidity. The emergency department diagnosis differed<br />
from the final discharge diagnosis in 4 <strong>of</strong> the 49 total patients. One such patient had received<br />
placebo <strong>and</strong> 3 had received high-dose morphine (no significant difference). There was no delay<br />
in outcome or time to treatment in any patient.<br />
Zoltie 8 demonstrated that 17/134 (12%) <strong>of</strong> those receiving opioids had altered physical<br />
signs. Of the 50 patients (out <strong>of</strong> 288 in the total opiate <strong>and</strong> placebo groups) whose signs changed<br />
during the evaluation, the most common change (n=32) was alteration in bowel sounds. The<br />
remaining 18 had altered sites <strong>of</strong> tenderness, in most cases a migration <strong>of</strong> a large region to a<br />
smaller, more precise area. In no case was the diagnosis altered by a change in physical signs. To<br />
the contrary, the correct diagnosis was facilitated in several cases, particularly in the 18 cases<br />
where the site <strong>of</strong> pain changed.<br />
Vermeulen et al 9 also found that the use <strong>of</strong> opioids did not change the appropriateness <strong>of</strong><br />
the surgeons’ decision making. Among female patients, the decision to operate was appropriate<br />
more <strong>of</strong>ten in the opioid group, but the difference between this group <strong>and</strong> the placebo group was<br />
not statistically significant. In male patients <strong>and</strong> overall, opiate analgesia did not influence the<br />
appropriateness <strong>of</strong> the decision. The appropriateness to discharge patients without surgery was<br />
100% in both groups. No patient who had left the hospital after 24 hours <strong>of</strong> observation without<br />
surgery was readmitted or operated on at another local hospital. The study also assessed the<br />
impact <strong>of</strong> analgesia on the accuracy <strong>of</strong> abdominal sonography. For diagnosis <strong>of</strong> appendicitis,<br />
ultrasound had lower sensitivity (71.1%) <strong>and</strong> higher specificity (65.2%) in the opioid group than<br />
in the placebo group, 80.6% <strong>and</strong> 53.8%, respectively.<br />
Similarly, Pace et al 10 found 3 diagnostic or management errors in each group (out <strong>of</strong> 35<br />
morphine <strong>and</strong> 36 control patients). The use <strong>of</strong> opioids did not alter the physicians’ ability to<br />
evaluate accurately <strong>and</strong> treat patients appropriately.<br />
Costs <strong>and</strong> Implementation<br />
The costs associated with implementing appropriate analgesic practice for patients with<br />
acute abdominal pain are limited to physician education programs <strong>and</strong> the cost <strong>of</strong> the analgesia<br />
<strong>and</strong> associated monitoring. There were no cost outcomes reported in any <strong>of</strong> the 5 studies.<br />
Comment<br />
From the available evidence, we conclude that appropriate use <strong>of</strong> analgesics in patients<br />
with acute abdominal pain effectively decreases pain <strong>and</strong> does not interfere with diagnosis or<br />
treatment. Recent surveys suggest many physicians believe conservative administration <strong>of</strong> pain<br />
medication does not interfere with diagnosis <strong>and</strong> treatment <strong>of</strong> patients with acute abdominal<br />
pain. Despite this recognition, the gap between underst<strong>and</strong>ing <strong>and</strong> practice remains large, <strong>and</strong><br />
abdominal pain is <strong>of</strong>ten undertreated.<br />
398
Table 37.1.1. R<strong>and</strong>omized controlled trials <strong>of</strong> analgesia in patients with acute abdominal<br />
pain*<br />
Study<br />
Zoltie, 1986 8<br />
Attard,<br />
1992 6<br />
Pace, 1996 10<br />
LoVecchio,<br />
1997 7<br />
Vermeulen,<br />
1999 9<br />
Study Participants;<br />
Intervention<br />
268 adults with acute<br />
abdominal pain<br />
admitted to a hospital in<br />
the UK; sublingual<br />
buprenorphine vs.<br />
placebo<br />
100 selected adults<br />
admitted with clinically<br />
significant abdominal<br />
pain; intramuscular<br />
papaveretum vs.<br />
placebo<br />
75 patients with acute<br />
abdominal pain at a US<br />
military emergency<br />
department; morphine<br />
49 adults with acute<br />
abdominal pain <strong>and</strong><br />
peritoneal signs (“acute<br />
abdomen”) admitted to<br />
the emergency<br />
department <strong>of</strong> a tertiary<br />
care hospital in NY;<br />
intravenous morphine<br />
vs. placebo<br />
340 adults with pain in<br />
the right lower part <strong>of</strong><br />
the abdomen at a<br />
university hospital<br />
emergency department;<br />
intravenous morphine<br />
vs. placebo<br />
Outcomes Results†<br />
Level 1 Pain better after 1 hour: 64/134 vs. 59/122 (p=NS). Only<br />
6/32 (19%) patients who received no tablet reported pain<br />
was better after 1 hour<br />
Change in physical signs after 1 hour: 22/134 (16%) vs.<br />
24/122 (20%); when site <strong>of</strong> tenderness changed, it usually<br />
was resolution <strong>of</strong> a large region to a smaller, precise area<br />
In no case was the diagnosis altered by a change in<br />
physical signs<br />
Level 1 Pain score: 3.1 vs. 8.3 (p
References<br />
1. Silen W. Cope’s early diagnosis <strong>of</strong> the acute abdomen, 17 th ed. New York: Oxford<br />
University Press; 1987.<br />
2. Wolfe JM, Lein DY, Lenkoski K, Smithline HA. Analgesic administration to patients with<br />
an acute abdomen: a survey <strong>of</strong> emergency medicine physicians. Am J Emerg Med<br />
2000;18:250-153.<br />
3. Silen W. Cope's Early Diagnosis <strong>of</strong> the Acute Abdomen. 19th ed. New York: Oxford<br />
University Press; 1996.<br />
4. Centers for Disease Control <strong>and</strong> Prevention. NCHS-FASTATS-Emergency Department<br />
Visits. Available at: http://www.cdc.gov/nchs/fastats/ervisits.htm. Accessed April 1, 2001.<br />
5. Graber MA, Ely JW, Clarke S, Kurtz S, Weir R. Informed consent <strong>and</strong> general surgeons’<br />
attitudes toward the use <strong>of</strong> pain medication in the acute abdomen. Am J Emerg Med<br />
1999;17:113-116.<br />
6. Attard AR, Corlett MJ, Kidner NJ, Leslie AP, Fraser IA. <strong>Safety</strong> <strong>of</strong> early pain relief for<br />
acute abdominal pain. BMJ 1992;305:554-556.<br />
7. LoVecchio F, Oster N, Sturmann K, Nelson LS, Flashner S, Finger R. The use <strong>of</strong><br />
analgesics in patients with acute abdominal pain. J Emerg Med 1997;15:775-779.<br />
8. Zoltie N, Cust MP. Analgesia in the acute abdomen. Ann R Coll Surg Engl 1986;68:209-<br />
210.<br />
9. Vermeulen B, Morabia A, Unger PF, Goehring C, Skljarov I, Terrier F. Acute appendicitis:<br />
influence <strong>of</strong> early pain relief on the accuracy <strong>of</strong> clinical <strong>and</strong> US findings in the decision to<br />
operate – a r<strong>and</strong>omized trial. Radiology 1999;210:639-643.<br />
10. Pace S, Burke TF. Intravenous morphine for early pain relief in patients with acute<br />
abdominal pain. Acad Emerg Med 1996;3:1086-1092.<br />
Subchapter 37.2. Acute Pain Services<br />
Background<br />
The concept <strong>of</strong> an acute pain service (APS) was first reported in 1988. Its genesis was the<br />
recognition <strong>of</strong> the problems wrought by inadequate postoperative pain management <strong>and</strong><br />
appreciation that acute pain may prolong recovery or precipitate complications. Over the past 15<br />
years, in the United States <strong>and</strong> worldwide, hospitals have created multidisciplinary acute pain<br />
services, with specially trained staff <strong>and</strong> resources geared toward providing up-to-date<br />
techniques <strong>and</strong> education. The APS creates a framework in which postoperative pain can be<br />
managed more effectively, hopefully leading to less discomfort <strong>and</strong> fewer postoperative<br />
complications.<br />
Practice Description<br />
An APS attempts to bridge the gap between physicians, nurses <strong>and</strong> patients to coordinate<br />
pain management. The roles <strong>of</strong> the APS are 1) educating patients, 2) educating nurses <strong>and</strong><br />
physicians, 3) selecting appropriate analgesic techniques for different situations, 4) preparing<br />
guidelines for different analgesic regimens, 5) helping to manage acute pain problems, <strong>and</strong> 6)<br />
performing quality control activities.<br />
Most acute pain services in the United States are anesthesiology-based. The<br />
comprehensive pain management teams usually consist <strong>of</strong> staff anesthesiologists, resident<br />
400
anesthesiologists, specially trained nurses, pharmacists <strong>and</strong> physiotherapists. Some services are<br />
nurse-based rather than anesthesia-based.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Approximately 50% <strong>of</strong> patients undergoing surgery do not receive adequate pain relief. 1<br />
Failure to appropriately treat pain stems from lack <strong>of</strong> knowledge <strong>and</strong> skills on the part <strong>of</strong> health<br />
care providers <strong>and</strong> those responsible for health care system management, <strong>and</strong> insufficient patient<br />
education. The “safety” problem targeted by the practice <strong>of</strong> implementing an APS is<br />
postoperative pain <strong>and</strong> morbidity.<br />
Opportunities for Impact<br />
Approximately 34-44% <strong>of</strong> hospitals in Europe 2,3 <strong>and</strong> most major institutions in the United<br />
States 4 have organized APSs. In general, few smaller hospitals have an APS. Therefore, there are<br />
many opportunities to institute acute pain services in hospitals.<br />
Study Designs<br />
Six articles were reviewed for this chapter. All are observational studies; one with a<br />
control, 5 without controls. All 6 studies looked at the intervention <strong>of</strong> an acute pain service in<br />
patients undergoing surgery (Table 37.2.1).<br />
Study Outcomes<br />
Three <strong>of</strong> the 6 studies assessed postoperative pain (Level 1). Two <strong>of</strong> the 6 studies<br />
assessed adverse effects <strong>and</strong> safety (Level 2) <strong>and</strong> one assessed knowledge <strong>and</strong> attitudes,<br />
perceived adequacy <strong>of</strong> patients’ pain relief <strong>and</strong> the effect on staff workload <strong>and</strong> relationships.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
All 3 studies that assessed postoperative pain scores found improvements. Bardiau et al 2<br />
showed that differences in pain score were most pronounced (around 50%) in patients<br />
undergoing vascular, maxill<strong>of</strong>acial, gynecologic, oral <strong>and</strong> urologic surgeries. Gould et al 5<br />
showed a reduction in median visual analog scores for pain during relaxation, movement <strong>and</strong><br />
deep inspiration. Tighe et al 6 showed a significant improvement in patient perception <strong>of</strong> pain<br />
relief after introduction <strong>of</strong> an APS.<br />
Schug <strong>and</strong> Torrie 7 found no complications resulting in sustained morbidity or mortality<br />
when anesthesiology-based APS provided postoperative pain relief. Potentially severe<br />
complications (without sequelae) occurred in 0.53% <strong>of</strong> patients. In one study by Tsui et al, 8 1.8%<br />
<strong>of</strong> patients developed respiratory complications (bradypnea, hypercapnia, oxygen desaturation),<br />
1.2% developed hypotension, <strong>and</strong> 28.8% <strong>and</strong> 15.1%, respectively, developed nausea <strong>and</strong><br />
vomiting. None suffered long-term sequelae.<br />
Although the postoperative setting is a logical place for acute pain services, they may<br />
also be useful in patients who experience pain as part <strong>of</strong> a disease process. Although used more<br />
for managing chronic conditions such as cancer <strong>and</strong> low back pain, acute pain services are also<br />
gaining popularity in treating hospitalized patients with pain due to a medical condition. There<br />
are no rigorous trials <strong>of</strong> APSs as they are used for medical patients.<br />
Potential for Harm<br />
Fragmentation <strong>of</strong> care (ie, lack <strong>of</strong> continuity between the anesthesiologist performing<br />
preoperative evaluation <strong>and</strong> anesthesiologist providing postoperative pain control, or<br />
401
fragmentation <strong>of</strong> care among multiple physicians) <strong>and</strong> decreased attention by the physician-<strong>of</strong>record<br />
may result in problems from the intervention. However, no studies have examined these<br />
concerns.<br />
Costs <strong>and</strong> Implementation<br />
Although none <strong>of</strong> the studies directly examined costs <strong>of</strong> implementing an acute pain<br />
service, one study estimated that an APS might be cost-effective. 6 Some data suggest that a<br />
nurse-based APS may be more cost-effective than an anesthesiologist-based APS, although there<br />
are no formal analyses <strong>of</strong> this supposition. 9<br />
Principal obstacles to implementing such acute pain services include financial<br />
constraints, the challenges <strong>of</strong> educating newly qualified doctors regarding pain management, <strong>and</strong><br />
the complexity <strong>of</strong> published guidelines. 9<br />
Comment<br />
Studies <strong>of</strong> APSs are mostly observational, measuring postoperative pain, adverse<br />
outcomes <strong>and</strong> staff knowledge <strong>and</strong> attitudes regarding its implementation. Although these studies<br />
indicate that acute pain services can improve postoperative pain without endangering patient<br />
safety, no formal recommendation can be made in the absence <strong>of</strong> high quality, systematic<br />
reviews <strong>of</strong> the benefits, costs <strong>and</strong> feasibility <strong>of</strong> implementing these services.<br />
402
Table 37.2.1. Studies <strong>of</strong> acute pain services in postoperative pain management*<br />
Study Setting<br />
1304 patients in the pre-APS<br />
inception phase <strong>and</strong> 671<br />
patients after its implementation<br />
undergoing various surgeries in<br />
a university teaching hospital 2<br />
2035 patients undergoing<br />
various surgical operations at a<br />
university hospital 5<br />
1518 patients undergoing<br />
various surgeries at a district<br />
general hospital 6<br />
2509 patients under APS care at<br />
a tertiary referral teaching<br />
hospital; 1759 received<br />
systemic analgesia, 590<br />
epidural; 160 other techniques 8<br />
3016 patients treated by an APS<br />
for postoperative pain 7<br />
48 staff members (36 nurses, 12<br />
house <strong>of</strong>ficers) working in two<br />
surgical units 10<br />
Study Design,<br />
Outcomes<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 3<br />
403<br />
Results<br />
* APS indicates acute pain service; CI, confidence interval.<br />
Significant reduction <strong>of</strong> all pain indicators<br />
after APS inception (p50%) in patients<br />
undergoing vascular, maxill<strong>of</strong>acial,<br />
gynecologic, urologic <strong>and</strong> oral surgeries<br />
Reduction in mean pain from 45 (95% CI:<br />
34-53) to 16 (95% CI: 10-20) after APS<br />
Significant reduction <strong>of</strong> pain (p
References<br />
1. AHCPR. Acute pain management: operative or medical procedures <strong>and</strong> trauma, part 1. Clin<br />
Pharm 1992;11:309-331.<br />
2. Bardiau FM, Braeckman MM, Seidel L, Albert A, Boogaerts JG. Effectiveness <strong>of</strong> an acute<br />
pain service inception in a general hospital. J Clin Anesth 1999;11:583-589.<br />
3. Hall PA, Bowden MI. Introducing an acute pain service. Br J Hosp Med 1996;55:15-17.<br />
4. Ready LB. How many acute pain services are there in the US <strong>and</strong> who is managing patientcontrolled<br />
analgesia? Anesth 1995;82:322.<br />
5. Gould TH, Crosby DL, Harmer M, Lloyd SM, Lunn JN, Rees GAD, Roberts DE, Webster<br />
JA. Policy for controlling pain after surgery: effect <strong>of</strong> sequential changes in management.<br />
BMJ 1992;305:1187-1193.<br />
6. Tighe SQ, Bie JA, Nelson RA, Skues MA. The acute pain service: effective or expensive<br />
care? Anaesth 1998;53:382-403.<br />
7. Schug SA, Torrie JJ. <strong>Safety</strong> assessment <strong>of</strong> postoperative pain management by an acute pain<br />
service. Pain 1993;55:387-391.<br />
8. Tsui SL, Irwin MG, Wong CML, Fung SKY, Hui TWC, Ng, KFJ, Chan WS, O’Reagan<br />
AM. An audit <strong>of</strong> the safety <strong>of</strong> an acute pain service. Anaesth 1997;52:1042-1047.<br />
9. Rawal N. Ten years <strong>of</strong> acute pain services – achievements <strong>and</strong> challenges. Reg Anesth Pain<br />
Med 1999;24:68-73.<br />
10. McLeod GA, Davies HTO, Colvin JR. Shaping attitudes to postoperative pain relief: the<br />
role <strong>of</strong> the acute pain team. J Pain Symptom Manage 1995;10:30-34.<br />
Subchapter 37.3. Prophylactic Antiemetics During <strong>Patient</strong>-controlled Analgesia Therapy<br />
Background<br />
Nausea <strong>and</strong> vomiting are common side effects <strong>of</strong> patient-controlled analgesia (PCA) with<br />
opioids. When severe, nausea <strong>and</strong> vomiting may limit a patient’s tolerance <strong>of</strong> PCA. Giving<br />
antiemetics prophylactically has been suggested as a way to prevent PCA-associated nausea <strong>and</strong><br />
vomiting.<br />
Practice Description<br />
No study has definitively determined the best prophylactic antiemetic in patientcontrolled<br />
analgesia. Prophylactic treatments that have been used include droperidol, 5-HT3<br />
receptor antagonists (ondansetron, tropisetron), clonidine, promethazine, hyoscine, prop<strong>of</strong>ol <strong>and</strong><br />
metoclopramide. Prophylactic antiemetics have been given both at the induction <strong>of</strong> anesthesia<br />
<strong>and</strong> at the end <strong>of</strong> surgery. A relatively new approach to minimizing nausea <strong>and</strong> vomiting<br />
associated with PCA involves the addition <strong>of</strong> an antiemetic to PCA, so that the patient receives<br />
both analgesic <strong>and</strong> antiemetic with each PCA dem<strong>and</strong> dose.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The incidence <strong>of</strong> postoperative nausea <strong>and</strong> vomiting has varied widely (from 8 to 92%)<br />
among studies. 1 There is no evidence that the incidence <strong>of</strong> nausea <strong>and</strong> vomiting with PCA is any<br />
different than that with intramuscular opioids. In patients receiving PCA, 57-90% report nausea<br />
<strong>and</strong> 27-40% report vomiting. There appears to be no clear advantage to using one opioid over<br />
any others in PCA in terms <strong>of</strong> postoperative emesis.<br />
404
Opportunities for Impact<br />
The degree to which prophylactic antiemetics are used in routine practice with PCA is<br />
unknown.<br />
Study Designs<br />
Tramer <strong>and</strong> Walder 2 conducted a systematic search for r<strong>and</strong>omized trials (MEDLINE,<br />
Embase, Cochrane library, reference lists, h<strong>and</strong>-searching, no language restriction) that<br />
compared prophylactic antiemetics with placebo or no treatment in patients receiving<br />
postoperative PCA with opioids. Their review identified 14 placebo-controlled trials with<br />
different regimens <strong>of</strong> droperidol, ondansetron, hyoscine transdermal therapeutic system,<br />
tropisetron, metoclopramide, prop<strong>of</strong>ol <strong>and</strong> promethazine. One PCA delivered tramadol; all<br />
others delivered morphine.<br />
Both relative risk <strong>and</strong> number needed to treat were calculated. To estimate the frequency<br />
<strong>of</strong> drug-related adverse effects, the relative risk <strong>and</strong> the number needed to harm were calculated.<br />
Study Outcomes<br />
The main end point for efficacy was prevention <strong>of</strong> emetic events (Level 1). The incidence<br />
<strong>of</strong> emetic events with active treatments (experimental event rate) <strong>and</strong> with placebo or no<br />
treatment (control event rate) was extracted. Nausea, vomiting, <strong>and</strong> “any emetic event” (nausea,<br />
vomiting, or nausea <strong>and</strong> vomiting) were extracted from each trial. Data on drug-related adverse<br />
effects (Level 2) were extracted as well.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Without antiemetic drugs, the incidence <strong>of</strong> nausea averaged 43% (range 22-80%),<br />
vomiting 55% (45-71%), <strong>and</strong> any emetic event 67% (54-87%). At 24 hours postoperatively, the<br />
cumulative incidence <strong>of</strong> nausea <strong>and</strong> vomiting in patients not receiving any antiemetic treatment<br />
added to their PCA-morphine was approximately 50%.<br />
The best-studied antiemetic was droperidol. It was added to morphine-PCA in 6 placebocontrolled<br />
trials involving 642 patients. Droperidol 0.017-0.17 mg/mg <strong>of</strong> morphine (0.5-11 mg/d<br />
droperidol) was significantly more effective (p=0.04) in preventing nausea than placebo, without<br />
evidence <strong>of</strong> dose-responsiveness. Compared with placebo, the number needed to treat with<br />
droperidol to prevent nausea was 2.7 (95% CI: 1.8-5.2) <strong>and</strong> to prevent vomiting was 3.2 (95%<br />
CI: 2.3-4.8).<br />
The second most frequently reported drugs were 5-HT3 receptor antagonists<br />
(ondansetron, tropisetron). Their effect on vomiting was satisfactory, with numbers needed to<br />
treat <strong>of</strong> approximately 5 compared with placebo. There was no evidence <strong>of</strong> any antinausea effect.<br />
Promising results were shown with some <strong>of</strong> the other interventions (clonidine,<br />
promethazine). However, the limited numbers <strong>of</strong> patients studied did not generate sufficient<br />
evidence to make a recommendation.<br />
405
Potential for Harm<br />
With placebo, the absolute risk <strong>of</strong> minor adverse events (sedation, drowsiness <strong>and</strong><br />
dizziness or anxiety, restlessness <strong>and</strong> agitation) was 0-20%. In those trials that reported adverse<br />
events with droperidol, doses <strong>of</strong> the antiemetic ranged from 1.2 mg to cumulative doses <strong>of</strong> 7.4<br />
mg. There were no obvious differences in the incidence <strong>of</strong> minor adverse effects compared with<br />
placebo with droperidol doses
Practice Description<br />
Non-pharmacologic interventions typically fall into 3 categories: health care information,<br />
skills teaching <strong>and</strong> psychosocial support. 2 Health care information includes information on<br />
preparation for surgery, timing <strong>of</strong> procedures, functions <strong>and</strong> roles <strong>of</strong> health care providers, selfcare<br />
responsibilities, <strong>and</strong> pain/discomfort information. Skills teaching includes coughing,<br />
breathing <strong>and</strong> bed exercises, hypnosis, cognitive reappraisal <strong>and</strong> relaxation exercises. Relaxation<br />
strategies include such techniques as the modified Jacobson method, Flaherty <strong>and</strong> Fitzpatrick<br />
jaw relaxation, relaxation tapes, tapes with structured breathing, muscle relaxation <strong>and</strong> pleasant<br />
imagery, <strong>and</strong> cognitive relaxation. Psychosocial support includes identifying <strong>and</strong> alleviating<br />
concerns, providing reassurance, problem solving with the patient, encouraging questions <strong>and</strong><br />
increasing frequency <strong>of</strong> support.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The safety problem targeted by the practice <strong>of</strong> non-pharmacologic interventions is<br />
postoperative pain. At least 40-50% <strong>of</strong> postoperative patients report inadequate pain relief,<br />
despite pharmacologic interventions. 3 Postoperative pain can have deleterious psychological <strong>and</strong><br />
physiologic consequences that contribute to patient discomfort <strong>and</strong> longer recovery periods, <strong>and</strong><br />
may compromise outcomes. 1 It also consumes greater health care resources.<br />
Opportunities for Impact<br />
Non-pharmacologic interventions are increasingly used in management <strong>of</strong> postoperative<br />
pain, although the exact proportion <strong>of</strong> patients receiving such interventions is unknown.<br />
Study Designs<br />
We identified 7 meta-analyses <strong>of</strong> studies evaluating the effectiveness <strong>of</strong> nonpharmacologic<br />
interventions on postoperative pain. Two <strong>of</strong> these meta-analyses evaluated<br />
various non-pharmacologic interventions for the management <strong>of</strong> acute pain. Devine 2 conducted a<br />
meta-analysis <strong>of</strong> 191 studies looking at psychoeducational care (including all 3 intervention<br />
categories above) for adult surgical patients. Sindhu 3 conducted a meta-analysis including 49<br />
r<strong>and</strong>omized controlled trials looking at non-pharmacologic nursing interventions (preoperative<br />
education/information, relaxation, music, imagery, bi<strong>of</strong>eedback, multidisciplinary approaches<br />
<strong>and</strong> others) for the management <strong>of</strong> acute pain. AHCPR 1 conducted a systematic search <strong>of</strong> nondrug<br />
intervention studies on postoperative pain management. One hundred forty studies were<br />
included <strong>and</strong> formal meta-analysis <strong>of</strong> 3 was performed. The American Society <strong>of</strong><br />
Anesthesiologists Task Force on Pain Management, Acute Pain Section 4 performed a systematic<br />
search for acute pain management in the perioperative setting. Two hundred thirty-three articles<br />
were used in the formal meta-analysis; the number <strong>of</strong> articles regarding education <strong>of</strong> patients was<br />
not reported. Good 5 performed a systematic search <strong>of</strong> trials evaluating the effects <strong>of</strong> relaxation<br />
<strong>and</strong> music on postoperative pain. Suls <strong>and</strong> Wan 6 performed a systematic search <strong>of</strong> trials<br />
evaluating the effects <strong>of</strong> sensory <strong>and</strong> procedural information on coping with stressful medical<br />
procedures <strong>and</strong> pain. Seers <strong>and</strong> Carroll 7 performed a systematic review <strong>of</strong> various relaxation<br />
techniques for postoperative pain relief. Table 37.4.1 lists the 7 articles <strong>and</strong> briefly describes<br />
their salient features.<br />
407
Study Outcomes<br />
All studies reported postoperative pain (Level 1) as a primary outcome measure. Other<br />
outcomes, such as recovery time, psychological distress, <strong>and</strong> opioid intake were reported in some<br />
cases (Level 1).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The extensive literature covered by the meta-analyses suggested beneficial effects <strong>of</strong><br />
psychoeducational care, education <strong>and</strong> instruction <strong>of</strong> patients, music <strong>and</strong> relaxation techniques<br />
(Table 39.5.1). The 2 meta-analyses that examined opioid intake failed to show an effect <strong>of</strong> nonpharmacologic<br />
interventions in reducing postoperative opioid consumption.<br />
Potential for Harm<br />
One study 7 reported no adverse events in any <strong>of</strong> the trials for any <strong>of</strong> the treatment or<br />
control groups. Otherwise, adverse consequences <strong>of</strong> non-pharmacologic interventions on<br />
postoperative pain management were not addressed.<br />
Costs <strong>and</strong> Implementation<br />
Direct information on costs <strong>of</strong> non-pharmacologic interventions was not reported in these<br />
7 studies. Devine <strong>and</strong> Cook 8 found that the beneficial impact on length <strong>of</strong> stay <strong>and</strong> medical<br />
complications rendered non-pharmacologic interventions cost-beneficial. Another review 2<br />
hypothesized that costs could potentially be decreased by non-pharmacologic interventions,<br />
since the length <strong>of</strong> hospital stay was shortened by an average <strong>of</strong> 1.5 days (11.5%).<br />
The most obvious direct cost <strong>of</strong> psychoeducational care is the increased staff time to<br />
provide these services. Based on average treatment duration <strong>of</strong> 42 minutes, a comprehensive<br />
version <strong>of</strong> the intervention would probably not take more than 1 hour per patient. 8 Less obvious<br />
direct costs might result from staff time to plan the protocol for psychoeducational care <strong>and</strong>/or to<br />
develop patient education materials, in-service programs or staff meetings to teach or review the<br />
protocol, printing or purchasing patient education materials, transporting patients to group<br />
teaching sessions, <strong>and</strong> staff time to document the level <strong>of</strong> care provided.<br />
Comment<br />
Effective treatment <strong>of</strong> postoperative pain continues to be a challenge. In addition to<br />
analgesia, non-pharmacologic interventions may provide some benefit in reducing postoperative<br />
pain. Clinicians have a wide range <strong>of</strong> options to consider when developing a comprehensive<br />
version <strong>of</strong> non-pharmacologic care appropriate for their patients. As such interventions are low<br />
risk <strong>and</strong> appeal to many patients, they should be explored in practice <strong>and</strong> further research.<br />
408
Table 37.4.1. Studies <strong>of</strong> non-pharmacologic interventions for postoperative pain<br />
Study setting; Practice<br />
Examined<br />
Adult surgical patients (# not<br />
stated), 92% in American hospitals<br />
(45% teaching, 43% general<br />
hospitals); health care information,<br />
skills teaching, psychosocial<br />
support 2<br />
Not reported; transcutaneous nerve<br />
stimulation, education/instruction,<br />
relaxation 1<br />
Not reported; education <strong>and</strong><br />
participation <strong>of</strong> patients <strong>and</strong><br />
families in pain control 4<br />
Adult pts who have undergone torso<br />
surgery (# not stated); relaxation<br />
techniques, music 5<br />
Not reported; sensory <strong>and</strong><br />
procedural information<br />
(explanations <strong>of</strong> what will be<br />
happening <strong>and</strong> how patients can<br />
expect to feel) 6<br />
362 patients undergoing fractured<br />
hip repair, removal <strong>of</strong> malignant<br />
skin lesions, major elective<br />
abdominal surgery, elective<br />
cholecystectomy, abdominal<br />
hysterectomy, femoral angiography;<br />
jaw relaxation, imagery, music,<br />
breathing, relaxation tapes 7<br />
3387 adult patients; education,<br />
relaxation, music, imagery,<br />
bi<strong>of</strong>eedback, multidisciplinary<br />
approach 3<br />
Study<br />
Design,<br />
Outcomes<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
Level 1-3 A,<br />
Level 1<br />
409<br />
Results*<br />
79-84% <strong>of</strong> studies found beneficial effects;<br />
Average effect size values were 0.43 for<br />
recovery, 0.38 for pain, 0.36 for psychological<br />
distress;<br />
Length <strong>of</strong> stay decreased 11.5%<br />
Simple/complex relaxation techniques,<br />
education/ instruction effective in reducing<br />
mild-moderate pain<br />
Education improves pain control, reduces<br />
adverse outcomes<br />
Total pain decreased in 10 <strong>of</strong> 13 studies;<br />
Sensory pain† reduction was reported in 6 <strong>of</strong> 12<br />
studies;<br />
Affective pain‡ decreased in 10 <strong>of</strong> 13 studies;<br />
Unidimensional pain decreased in 4 <strong>of</strong> 7 studies;<br />
Observed pain decreased in 4 <strong>of</strong> 4 studies<br />
Combination <strong>of</strong> procedural <strong>and</strong> sensory<br />
preparation significantly better than control on<br />
all measures; effect sizes with combination<br />
larger than with either type <strong>of</strong> information alone<br />
3 <strong>of</strong> 7 studies showed significant decrease in<br />
pain sensation/pain distress in those who had<br />
relaxation; 1 out <strong>of</strong> 5 trials showed significant<br />
improvement in psychological outcomes; less<br />
anxiety in relaxation group<br />
Effect sizes ranged from 2.25 to 1.78; strong<br />
heterogeneity<br />
* Effect size is the st<strong>and</strong>ardized difference between the control <strong>and</strong> experimental groups regarding the<br />
measure <strong>of</strong> interest in each study. Positive values indicate benefit with the intervention.<br />
† Sensory pain refers to ability to discriminate where pain is occurring <strong>and</strong> respond appropriately.<br />
‡ Affective pain is the sensation <strong>of</strong> pain as something unpleasant <strong>and</strong> to be avoided.
References<br />
1. AHCPR Pain Management Guideline Panel. Clinicians’ quick reference guide to<br />
postoperative pain management in adults. J Pain Symptom Manage 1992;7:214-228.<br />
2. Devine EC. Effects <strong>of</strong> psychoeducational care for adult surgical patients: a meta-analysis <strong>of</strong><br />
191 studies. <strong>Patient</strong> Educ Couns 1992;19:129-142.<br />
3. Sindhu F. Are non-pharmacological nursing interventions for the management <strong>of</strong> pain<br />
effective? – a meta-analysis. J Adv Nurs. 1996;24:1152-1159.<br />
4. A Report by the American Society <strong>of</strong> Anesthesiologists Task Force on Pain Management,<br />
Acute Pain Section. Practice guidelines for acute pain management in the perioperative<br />
setting. Anesth 1995;82:1071-1081.<br />
5. Good M. Effects <strong>of</strong> relaxation <strong>and</strong> music on postoperative pain: a review. J Adv Nurs.<br />
1996;24:905-914.<br />
6. Suls J, Wan CK. Effects <strong>of</strong> sensory <strong>and</strong> procedural information on coping with stressful<br />
medical procedures <strong>and</strong> pain: a meta-analysis. J Consult Clin Psychol 1989;57:372-33379.<br />
7. Seers K, Carroll D. Relaxation techniques for acute pain management: a systematic review.<br />
J Adv Nurs 1998;27:466-475.<br />
8. Devine EC, Cook TD. Clinical <strong>and</strong> cost-saving effects <strong>of</strong> psychoeducational interventions<br />
with surgical patients: a meta-analysis. Res Nurs Health 1986;9:89-105.<br />
410
Section F. Organization, Structure, <strong>and</strong> Culture<br />
Chapter 38. "Closed" Intensive Care Units <strong>and</strong> Other Models <strong>of</strong> Care for <strong>Critical</strong>ly Ill<br />
<strong>Patient</strong>s<br />
Chapter 39. Nurse Staffing, Models <strong>of</strong> Care Delivery, <strong>and</strong> Interventions<br />
Chapter 40. Promoting a Culture <strong>of</strong> <strong>Safety</strong><br />
411
412
Chapter 38. “Closed” Intensive Care Units <strong>and</strong> Other Models <strong>of</strong> Care for<br />
<strong>Critical</strong>ly Ill <strong>Patient</strong>s<br />
Jeffrey M. Rothschild, MD, MPH<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
<strong>Patient</strong>s in the intensive care unit (ICU) require complex care relating to a broad range <strong>of</strong><br />
acute illnesses <strong>and</strong> pre-existing conditions. The innate complexity <strong>of</strong> the ICU makes<br />
organizational structuring <strong>of</strong> care an attractive quality measure <strong>and</strong> a target for performance<br />
improvement strategies. In other words, organizational features relating to medical <strong>and</strong> nursing<br />
leadership, communication <strong>and</strong> collaboration among providers, <strong>and</strong> approaches to problemsolving<br />
1 may capture the quality <strong>of</strong> ICU care more comprehensively than do practices related to<br />
specific processes <strong>of</strong> care. 2<br />
Most features <strong>of</strong> ICU organization do not exert a demonstrable impact on clinical<br />
outcomes such as morbidity <strong>and</strong> mortality. 3 While hard clinical outcomes may not represent the<br />
most appropriate measure <strong>of</strong> success for many organizational features, the role <strong>of</strong> “intensivists”<br />
(specialists in critical care medicine) in managing ICU patients has shown a beneficial impact on<br />
patient outcomes in a number <strong>of</strong> studies. For this reason, the Leapfrog Group, representing<br />
Fortune 500 corporations <strong>and</strong> other large health care purchasers, has identified staffing ICUs<br />
with intensivists as one <strong>of</strong> three recommended hospital safety initiatives for its 2000 purchasing<br />
principles (see also Chapter 55). 4<br />
In this chapter, we review the benefits <strong>of</strong> full-time intensivists <strong>and</strong> the impact <strong>of</strong> “closed<br />
ICUs” (defined below) on patient outcomes. Much <strong>of</strong> this literature makes no distinction<br />
between improved outcomes in general <strong>and</strong> decreased harm in particular. However, given the<br />
high mortality 5 <strong>and</strong> complication rates 6-8 observed in ICUs, it seems reasonable to consider<br />
global interventions such as organizational changes as patient safety practices.<br />
Practice Description<br />
The following practice definitions are synthesized from studies reviewed for this chapter.<br />
For all <strong>of</strong> these models, the term “intensivist” refers to a physician with primary training in<br />
medicine, surgery, anesthesiology or pediatrics followed by 2-3 years <strong>of</strong> critical care medicine<br />
(CCM) training.<br />
Open ICU model—An ICU in which patients are admitted under the care <strong>of</strong> an internist,<br />
family physician, surgeon or other primary attending <strong>of</strong> record, with intensivists available<br />
providing expertise via elective consultation. Intensivists may play a de facto primary role in the<br />
management <strong>of</strong> some patients, but only within the discretion <strong>of</strong> the attending-<strong>of</strong>-record.<br />
Intensivist Co-management—An open ICU model in which all patients receive<br />
m<strong>and</strong>atory consultation from an intensivist. The internist, family physician, or surgeon remains a<br />
co-attending-<strong>of</strong>-record with intensivists collaborating in the management <strong>of</strong> all ICU patients.<br />
Closed ICU model—An ICU in which patients admitted to the ICU are transferred to the<br />
care <strong>of</strong> an intensivist assigned to the ICU on a full-time basis. Generally, patients are accepted to<br />
the ICU only after approval/evaluation by the intensivist. For periods typically ranging from one<br />
week to one month at a time, the intensivist’s clinical duties predominantly consist <strong>of</strong> caring for<br />
patients in the ICU, with no concurrent outpatient responsibilities.<br />
413
Mixed ICU models—In practice, the above models overlap to a considerable extent.<br />
Thus, some studies avoid attempting to characterize ICUs in terms <strong>of</strong> these models <strong>and</strong> focus<br />
instead on the level <strong>of</strong> involvement <strong>of</strong> intensivists in patient care regardless <strong>of</strong> the organizational<br />
model. This involvement may consist <strong>of</strong> daily ICU rounds by an intensivist (thus including<br />
“closed model ICUs” <strong>and</strong> “intensivist comanagement”), ICU directorship by an intensivist<br />
(possibly including examples <strong>of</strong> all 3 models above), or simply the presence <strong>of</strong> a full-time<br />
intensivist in the ICU (also including examples <strong>of</strong> all 3 models.)<br />
Intensivist models—ICU management may include all <strong>of</strong> these models. These models are<br />
contrasted with the open ICU model, in which an intensivist generally does not participate in the<br />
direct care <strong>of</strong> a significant proportion <strong>of</strong> the ICU patients.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
ICUs comprise approximately 10% <strong>of</strong> acute care hospital beds. 9 The number <strong>of</strong> annual<br />
ICU admissions in the United State is estimated to be 4.4 million patients. 10 Due to an aging<br />
population <strong>and</strong> the increasing acuity <strong>of</strong> illness <strong>of</strong> hospitalized patients, both the total number <strong>of</strong><br />
ICU patients <strong>and</strong> their proportional share <strong>of</strong> hospital admissions overall are expected to grow. 11<br />
ICU patients have, on average, mortality rates between 12 <strong>and</strong> 17%. 25 Overall,<br />
approximately 500,000 ICU patients die annually in the United States. A recent review estimated<br />
that this mortality could be reduced by 15 to 60% using an intensivist model <strong>of</strong> ICU<br />
management. 12<br />
Young <strong>and</strong> Birkmeyer have provided estimates <strong>of</strong> the relative reduction in annual ICU<br />
mortalities resulting from conversion <strong>of</strong> all urban ICUs to an intensivist model <strong>of</strong> management<br />
model. 10 Using conservative estimates for current ICU mortality rates <strong>of</strong> 12%, <strong>and</strong> estimating<br />
that 85% <strong>of</strong> urban ICUs are not currently intensivist-managed, the authors calculated that<br />
approximately 360,000 patients die annually in urban ICUs without intensivists. A conservative<br />
projection <strong>of</strong> a 15% relative reduction in mortality resulting from intensivist-managed ICUs<br />
yields a predicted annual saving <strong>of</strong> nearly 54,000 lives.<br />
By only measuring ICU mortality rates, this analysis may underestimate the importance<br />
<strong>of</strong> intensivist-managed ICUs. In addition to mortality, other quality <strong>of</strong> care outcome measures<br />
that might be improved by intensivists include rates <strong>of</strong> ICU complications, inappropriate ICU<br />
utilization, patient suffering, appropriate end-<strong>of</strong>-life palliative care, <strong>and</strong> futile care.<br />
Opportunities for Impact<br />
Currently, a minority <strong>of</strong> ICUs in the United States utilizes the intensivist model <strong>of</strong> ICU<br />
management. 13 Intensivists are even less frequently found in non-teaching <strong>and</strong> rural hospitals.<br />
The potential impact <strong>of</strong> the intensivist model is far-reaching.<br />
Study Designs<br />
Among 14 studies abstracted for this chapter, 2 were systematic reviews <strong>and</strong> 12 were<br />
original studies. One systematic review is an abstract that has not yet appeared in journal form<br />
<strong>and</strong> does not provide cited references. 12 The other systematic review evaluated 8 references, all<br />
<strong>of</strong> which are included in this chapter. 10 An additional 4 studies absent from the systematic review<br />
are included here. These 4 studies include 2 abstracts that were published after the 1999<br />
systematic review, 14,15 <strong>and</strong> 2 studies <strong>of</strong> pediatric ICUs with intensivists. 16,17<br />
Among the original studies, 6 incorporated historical controls <strong>and</strong> 5 used a crosssectional<br />
approach. One study 18 had both historical <strong>and</strong> cross-sectional components. The original<br />
studies include 4 studies <strong>of</strong> adult medical ICUs, 6 studies <strong>of</strong> adult surgical ICUs <strong>and</strong> 2 studies <strong>of</strong><br />
414
pediatric multidisciplinary ICUs. Intensivist models used by the studies cited for this review<br />
include 4 closed ICUs, 4 mixed ICUs, 3 ICUs with intensivist comanagement <strong>and</strong> one open ICU.<br />
Several studies were excluded, including abstracts with insufficient data, 19-25 unclear<br />
distinctions in patient management between control groups <strong>and</strong> intervention (intensivist<br />
managed) groups, 26,27 intensivist models that may have important roles in future practice (eg,<br />
telemedicine consultation with remote management) but are not yet widely available 28,29 <strong>and</strong><br />
considerably older studies. 30<br />
Study Outcomes<br />
Required outcomes <strong>of</strong> interest in studies chosen for this chapter were ICU mortality,<br />
overall in-hospital mortality, or both. Some studies also included morbidity outcomes, adverse<br />
events <strong>and</strong> resource utilization (eg, length <strong>of</strong> ICU <strong>and</strong> hospital stay), levels <strong>of</strong> patient acuity or<br />
severity <strong>of</strong> illness (ICU utilization) <strong>and</strong> levels <strong>of</strong> high-intensity intervention usage. Studies<br />
addressing the impact <strong>of</strong> intensivist ICU management on resource utilization without mortality<br />
or outcome data were excluded. There are no data regarding the impact <strong>of</strong> intensivists.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
As shown in Table 38.1, most <strong>of</strong> the studies report a decrease in unadjusted in-hospital<br />
mortality <strong>and</strong>/or ICU mortality, although this decrease did not reach statistical significance in 3<br />
<strong>of</strong> the 14 studies. 16,18,31 One study found a statistically insignificant increase in the unadjusted<br />
mortality rates associated with the intensivist model ICU. 32 This study also found that the ratio <strong>of</strong><br />
expected-to-actual mortality was reduced in the intensivist-model ICUs. This finding was<br />
associated with a higher severity <strong>of</strong> illness scores in the intensivist-model ICU population. A<br />
similar finding <strong>of</strong> significantly improved outcomes after adjusting for severity <strong>of</strong> illness <strong>and</strong><br />
comparing expected-to-actual mortality rates was demonstrated in one pediatric study. 16 Overall,<br />
the relative risk reduction for ICU mortality ranges from 29% to 58%. The relative risk reduction<br />
for overall hospital mortality is 23% to 50%. These results are consistent with those <strong>of</strong> a<br />
previous systematic review that found a 15% to 65% reduction in mortality rates in intensivistmanaged<br />
ICUs. 10<br />
Data concerning long-term survival (6 <strong>and</strong> 12 months) for patients cared for in ICUs with<br />
<strong>and</strong> without intensivist management is not available. Differences in outcomes between closed<br />
ICUs, mixed ICU models <strong>and</strong> co-managed ICUs are difficult to assess. Studies that have<br />
addressed conversion from an open to a closed model did not utilize full-time intensivists in the<br />
open model study phases. 18,32-34 Therefore it is not clear to what extent improved patient<br />
outcomes resulted only from changes in intensivists’ direct patient care <strong>and</strong> supervision.<br />
The observational studies evaluating these practices suffer from 2 major limitations. Half<br />
<strong>of</strong> the studies retrospectively compared post-implementation outcomes with those during an<br />
historical control period. Because none <strong>of</strong> these studies included a similar comparison for a<br />
control unit that remained open in both time periods, we lack information on secular trends in<br />
ICU outcomes during the time periods evaluated. The other major limitation associated with<br />
comparing mortality rates for ICU patients relates to differences in ICU admission <strong>and</strong> discharge<br />
criteria under different organizational models. Under the intensivist model, patients are generally<br />
accepted to the ICU only after approval/evaluation by the intensivist. Thus, conversion to an<br />
intensivist model ICU may bring about changes in the ICU patient population that are<br />
incompletely captured by risk-adjustment models <strong>and</strong> confound comparisons <strong>of</strong> mortality rates.<br />
Moreover, these changes in ICU admitting practice may exert contradictory effects. For example,<br />
an intensivist model ICU may result in fewer ICU admissions for patients with dismal<br />
415
prognoses, <strong>and</strong> less futile care for patients already in the ICU. On the other h<strong>and</strong>, intensivistmanaged<br />
ICUs with stricter admission <strong>and</strong> discharge criteria may result in a greater overall<br />
acuity <strong>of</strong> illness for the ICU patients <strong>and</strong> therefore higher mortality rates.<br />
Potential for Harm<br />
The potential for harm resulting from intensivist management is unclear. Concerns raised<br />
in the literature about intensivist-managed ICUs include the loss <strong>of</strong> continuity <strong>of</strong> care by primary<br />
care physicians, insufficient patient-specific knowledge by the intensivist, 35 reduced use <strong>of</strong><br />
necessary sub-specialist consultations, <strong>and</strong> inadequate CCM training <strong>of</strong> residents who formerly<br />
managed their own ICU patients.<br />
Perhaps more worrisome is the impact that adoption <strong>of</strong> this practice would have on<br />
physician staffing <strong>and</strong> workforce requirements. Without a substantial increase in the numbers <strong>of</strong><br />
physicians trained in CCM, projected increases in the ICU patient population over the next 30<br />
years will result in a significant shortfall in the intensivist workforce. 11<br />
Costs <strong>and</strong> Implementation<br />
These studies did not address the incremental costs associated with implementation <strong>of</strong><br />
full-time intensivists. Several studies have analyzed resource utilization <strong>and</strong> length <strong>of</strong> stay<br />
associated with intensivist-managed ICUs. 13,16,18,19,29,31,32,36 The results <strong>of</strong> these studies are<br />
variable with respect to costs. Some demonstrate a decrease in ICU expenses. Others found<br />
increased costs, likely due to the increased use <strong>of</strong> expensive technologies. Still others show little<br />
overall cost differential. The cost-effectiveness <strong>and</strong> cost-benefit <strong>of</strong> an intensivist-model ICU<br />
requires further study.<br />
Comment<br />
Outcomes research in critical care is particularly challenging for several reasons. It<br />
typically relies on observational outcomes studies, <strong>and</strong> must account for the diversity <strong>and</strong><br />
complexity <strong>of</strong> variables measured <strong>and</strong> controlled for, such as patient-based, disease-based,<br />
provider-based <strong>and</strong> therapy-based variables. Despite these challenges <strong>and</strong> limitations, the<br />
literature fairly clearly shows that intensivists favorably impact ICU patient outcomes. What<br />
remains unclear is which intensivist model to recommend—intensivist consultation versus<br />
intensivist co-management versus closed ICUs. Also, we do not know the degree to which the<br />
choice among these models depends on intensivist background – ie, medicine, anesthesiology or<br />
surgery. Finally, because the mechanism <strong>of</strong> the benefit <strong>of</strong> intensivist models is unknown, the<br />
degree to which this benefit can be captured by other changes in practice (eg, adoption <strong>of</strong> certain<br />
evidence-based processes <strong>of</strong> ICU care) remains unclear.<br />
The major incentive for clarifying these issues concerns the implications for staffing<br />
ICUs in the future. While the evidence supports the beneficial role <strong>of</strong> full-time intensivists, the<br />
current number <strong>of</strong> trainees is insufficient to keep pace with the expected increase in the number<br />
<strong>of</strong> ICU patients. 11 Until we are able to sufficiently increase the size <strong>and</strong> number <strong>of</strong> CCM training<br />
programs for physician specialists, complementary solutions for meeting critical care<br />
management dem<strong>and</strong>s should be considered. These might include incorporating physicianextenders<br />
such as nurse practitioners <strong>and</strong> physician assistants with specialized critical care<br />
training, increased participation by hospitalists in care <strong>of</strong> ICU patients, 37 regionalization <strong>of</strong><br />
critical care services, 38 or providing innovative methods to extend intensivists’ expertise to<br />
remote sites through telemedicine consultations. 28 The latter practice seems particularly<br />
promising—a recent time series cohort study found an approximately 33% decrease in severity-<br />
416
adjusted hospital mortality <strong>and</strong> a nearly 50% decrease in ICU complications when a technologyenabled<br />
remote ICU management program was instituted in a community-based ICU. 28<br />
417
Table 38.1. Intensivist management in the care <strong>of</strong> critically ill patients*<br />
Study Setting Study<br />
Year<br />
Closed ICU Model<br />
Tertiary care, urban, teaching<br />
hospital; patients with septic<br />
shock; historical control 33<br />
Teaching hospitals (n=2); two<br />
study designs using historical<br />
<strong>and</strong> concurrent controls 18<br />
Tertiary care, urban, teaching<br />
hospital; historical control 32<br />
Tertiary care, urban, teaching<br />
hospital; historical control 34<br />
Mixed ICU models<br />
ICUs (n=16) with different<br />
characteristics; cross-sectional 16<br />
ICUs (n=39) with different<br />
characteristics; cross-sectional.<br />
<strong>Patient</strong>s with abdominal aortic<br />
surgery 38<br />
ICUs (n=31) with different<br />
characteristics; cross-sectional.<br />
<strong>Patient</strong>s with esophageal<br />
resection 14<br />
ICUs (n=39) with different<br />
characteristics; cross-sectional.<br />
<strong>Patient</strong>s with hepatic resection 15<br />
Community teaching hospital;<br />
historical control 40<br />
Co-managed ICUs<br />
Tertiary care ICU in a teaching<br />
children’s hospital 16<br />
Tertiary care, Canadian teaching<br />
hospital; historical control 39<br />
Tertiary care, urban, teaching<br />
hospital; cross-sectional<br />
comparison (concurrent<br />
control) 31<br />
1982-<br />
1984<br />
1992-<br />
1993<br />
1993-<br />
1994<br />
1995-<br />
1996<br />
1989-<br />
1992<br />
1994-<br />
1996<br />
1994-<br />
1998<br />
1994-<br />
1998<br />
1992-<br />
1994<br />
1983-<br />
1984<br />
1984-<br />
1986<br />
1994-<br />
1995<br />
ICU Type Study<br />
Design,<br />
Outcomes<br />
MICU Level 3,<br />
Level 1<br />
MICU Level 3,<br />
Level 1<br />
MICU Level 3,<br />
Level 1<br />
SICU Level 3,<br />
Level 1<br />
Pediatric<br />
MICU<br />
SICU<br />
Level 3,<br />
Level 1<br />
SICU Level 3,<br />
Level 1<br />
SICU Level 3,<br />
Level 1<br />
SICU Level 3,<br />
Level 1<br />
MICU Level 3,<br />
Level 1<br />
Pediatric<br />
MICU<br />
SICU<br />
Level 3,<br />
Level 3<br />
SICU Level 3,<br />
Level 1<br />
SICU Level 3,<br />
Level 1<br />
418<br />
Intensivist<br />
Intervention<br />
Mortality Relative Risk<br />
Reduction (%)<br />
ICU Hospital<br />
Closed NA 23<br />
Closed NA Retrospective:<br />
19 (p=NS)<br />
Prospective:<br />
26 (p=NS)<br />
Closed NA -38 (p=NS)†<br />
0/E 13‡<br />
Closed 58 50§<br />
Mixed RRR 25<br />
OR 1.5**<br />
NA<br />
Mixed NA OR 3.0§§<br />
Mixed NA RRR 73<br />
OR 3.5**<br />
Mixed NA RRR 81<br />
OR 3.8**<br />
Open 29 28<br />
Co-manage 48 (p=NS) NA<br />
Co-manage 52 31<br />
Co-manage NA 32 (p=NS)<br />
* ICU indicates intensive care unit; MICU, medicalintensive care unit; Mixed, mixed intensivist model (including daily<br />
ICU rounds by an intensivist, the presence <strong>of</strong> a full-time intensivist, open units with comanagement <strong>and</strong> closed units<br />
with m<strong>and</strong>atory consultations or only intensivist management); NA, not available as outcome (was not evaluated); NS,<br />
not stastically significant; <strong>and</strong> SICU, surgical intensive care unit.<br />
† Negative value indicates an increase in relative risk <strong>of</strong> mortality.<br />
‡ O/E is observed to expected mortality ratio based risk adjustment<br />
§ Hospital mortality measured 30-days after discharge
RRR is the unadjusted mortality relative risk reduction<br />
** OR is the adjusted odds ratio <strong>of</strong> increased mortality associated without an intensivist model.<br />
419
References<br />
1. Zimmerman JE, Shortell SM, Rousseau DM, Duffy J, Gillies RR, Knaus WA, et al.<br />
Improving intensive care: observations based on organizational case studies in nine<br />
intensive care units: a prospective, multicenter study. Crit Care Med. 1993;21:1443-1451.<br />
2. Shojania KG, Showstack J, Wachter RM. Assessing hospital quality: a review for<br />
clinicians. Eff Clin Pract. 2001;4:82-90.<br />
3. Mitchell PH, Shortell SM. Adverse outcomes <strong>and</strong> variations in organization <strong>of</strong> care<br />
delivery. Med Care. 1997;35(11 Suppl):NS19-NS32.<br />
4. Milstein A, Galvin R, Delbanco S, Salber P, Buck C. Improving the <strong>Safety</strong> <strong>of</strong> Health Care:<br />
The Leapfrog Initiative. Eff Clin Pract. 2000;3:313-316.<br />
5. Knaus WA, Wagner DP, Zimmerman JE, Draper EA. Variations in mortality <strong>and</strong> length <strong>of</strong><br />
stay in intensive care units. Ann Intern Med. 1993;118:753-761.<br />
6. Rubins H, Moskowitz M. Complications <strong>of</strong> Care in a <strong>Medical</strong> Intensive Care Unit. J Gen<br />
Intern Med. 1990;5:104-109.<br />
7. Giraud T, Dhainaut JF, Vaxelaire JF, Joseph T, Journois D, Bleichner G, et al. Iatrogenic<br />
complications in adult intensive care units: a prospective two-center study. Crit Care Med.<br />
1993;21:40-51.<br />
8. Ferraris VA, Propp ME. Outcome in critical care patients: a multivariate study. Crit Care<br />
Med. 1992;20:967-976.<br />
9. Groeger JS, Guntupalli KK, Strosberg M, Halpern N, Raphaely RC, Cerra F, et al.<br />
Descriptive analysis <strong>of</strong> critical care units in the United States: patient characteristics <strong>and</strong><br />
intensive care unit utilization. Crit Care Med. 1993;21:279-291.<br />
10. Young M, Birkmeyer J. Potential reduction in mortality rates using an intensivist model to<br />
manage intensive care units. Eff Clin Pract. 2000;3:284-289.<br />
11. Angus DC, MB C, Kelley MA, MD, Schmitz RJ, PhD, et al. Current <strong>and</strong> projected<br />
workforce requirements for care <strong>of</strong> the critically ill <strong>and</strong> patients with pulmonary disease:<br />
can we meet the requirements <strong>of</strong> an aging population? JAMA. 2000;284:2762-7270.<br />
12. Pronovost PJ, Young T, Dorman T, Robinson K, Angus D. Association between ICU<br />
physician staffing <strong>and</strong> outcomes: A systematic review. Crit.Care Med. 1999;27:A43.<br />
Abstract.<br />
13. Mallick R, Strosberg M, Lambrinos J, Groeger J. The intensive care unit medical director<br />
as manager: impact on performance. Med Care. 1995;33:611-624.<br />
14. Dimick J, Pronovost P, Heitmiller R, Lipsett P. ICU physician staffing is associated with<br />
decreased length <strong>of</strong> stay, hospital cost, <strong>and</strong> complications after esophageal resection. Paper<br />
presented at: Surgical Forum, American College <strong>of</strong> Surgeons, Chicago, Ill; v LI, p 493-495<br />
Oct 22, 2000.<br />
15. Dimick J, Pronovost P, Lipsett P. The effect <strong>of</strong> ICU physician staffing <strong>and</strong> hospital volume<br />
on outcomes after hepatic resection. Crit.Care Med. 2000;28:A77 Abstract.<br />
16. Pollack MM, Katz R, Ruttimann UE, Getson PR. Improving the outcome <strong>and</strong> efficiency <strong>of</strong><br />
intensive care: the impact <strong>of</strong> an intensivist. Crit Care Med. 1988;16:11-17.<br />
17. Pollack MM, MD, Cuerdon TT, PhD, Patel KM, Ruttimann UE, et al. Impact <strong>of</strong> quality-<strong>of</strong>care<br />
factors on pediatric intensive care unit mortality. JAMA. 1994;272:941-946.<br />
18. Multz AS, Chalfin DB, Samson IM, Dantzker DR, Fein AM, Steinberg HN, et al. A<br />
"closed" medical intensive care unit (MICU) improves resource utilization when compared<br />
with an "open" MICU. Am J Respir Crit Care Med. 1998;157:1468-1473.<br />
420
19. DiCosmo BFM. Addition <strong>of</strong> an intensivist improves ICU outcomes in a non-teaching<br />
community hospital. Chest. 1999;116:238S.<br />
20. Jacobs M, Hussain E, Hanna A, Ruskin G, Weiss S, Skrzpiec W. Improving the outcome<br />
<strong>and</strong> efficiency <strong>of</strong> surgical intensive care:The impact <strong>of</strong> full time medical intensivists.<br />
Chest. 1998;114:276S-277S.<br />
21. Render ML, Deddens JA, Thomas B, Wexler LF, Rouan GW. Decreased mortality among<br />
patients cared for in a closed intensive care unit setting. J Gen Intern Med. 1998;13 S 1:19.<br />
22. Marini C, Nathan I, Ritter G, Rivera L, Jurkiewicz A, Cohen J. The impact <strong>of</strong> full-time<br />
surgical intensivists on ICU utilization <strong>and</strong> mortality. Crit Care Med. 1995;23(Suppl 1):<br />
A235.<br />
23. Cowen J, Matchett S, Kaufman J, Baker K, Wasser T. Progressive reduction in severityadjusted<br />
mortality after implementation <strong>of</strong> a critical care program. Crit Care Med.<br />
1999;27(Suppl 1):35A.<br />
24. Tang G, Kuo H. Effect <strong>of</strong> a full time critical care specialist on ICU mortality. Crit Care<br />
Med. 1996;24:A37.<br />
25. Al-Asadi L, Dellinger R, Deutch J, Nathan S. Clinical impact <strong>of</strong> closed versus open<br />
provider care in a medical intensive care unit. American Journal <strong>of</strong> Respiratory & Crit<br />
Care Med. 1996;153:A360.<br />
26. Cole L, Bellomo R, Silvester W, Reeves J. A prospective, multicenter study <strong>of</strong> the<br />
epidemiology,management, <strong>and</strong> outcome <strong>of</strong> severe acute renal failure in a "closed" ICU<br />
system. Am J Resp Crit Care Med. 2000;162:191-196.<br />
27. Blunt M, Burchett K. Out-<strong>of</strong>-hours consultant cover <strong>and</strong> case-mix-adjusted mortality in<br />
intensive care. Lancet. 2001;356:735-736.<br />
28. Rosenfeld BA, MD FCCM, Dorman TM, Breslow MJ, Pronovost PM, PhD, Jenckes MM,<br />
et al. Intensive care unit telemedicine: Alternate paradigm for providing continuous<br />
intensivist care. Crit Care Med. 2000;28:3925-3931.<br />
29. Carlson RW, Weil<strong>and</strong> DE, Srivathsan K. Does a full-time, 24-hour intensivist improve care<br />
<strong>and</strong> efficiency? Crit Care Clin. 1996;12:525-551.<br />
30. Li TC, Phillips MC, Shaw L, Cook EF, Natanson C, Goldman L. On-site physician staffing<br />
in a community hospital intensive care unit. Impact on test <strong>and</strong> procedure use <strong>and</strong> on<br />
patient outcome. JAMA. 1984;252:2023-2027.<br />
31. Hanson CW, 3rd, Deutschman CS, Anderson HL, 3rd, Reilly PM, Behringer EC, Schwab<br />
CW, et al. Effects <strong>of</strong> an organized critical care service on outcomes <strong>and</strong> resource<br />
utilization: a cohort study. Crit Care Med. 1999;27:270-274.<br />
32. Carson SS, MD, Stocking CP, Podsadecki TM, Christenson JM, Pohlman AM, et al.<br />
Effects <strong>of</strong> organizational change in the medical intensive care unit <strong>of</strong> a teaching hospital: a<br />
comparison <strong>of</strong> 'open' <strong>and</strong> 'closed' formats. JAMA. 1996;276:322-328.<br />
33. Reynolds H, Haupt M, Thill-Baharozian M, Carlson R. Impact <strong>of</strong> critical care physician<br />
staffing on patients with septic shock in a university hospital medical intensive care unit.<br />
JAMA. 1988;260:3446-3450.<br />
34. Ghorra SM, Reinert SE, Ci<strong>of</strong>fi WM, Buczko GM, Simms H, Hank MD. <strong>Analysis</strong> <strong>of</strong> the<br />
effect <strong>of</strong> conversion from open to closed surgical intensive care unit. Ann Surg.<br />
1999;229:163-171.<br />
35. Trunkey D. An unacceptable concept. Arch Surg. 1999;229:172-173.<br />
36. Dicosmo B. Strict ICU admission/discharge criteria result in decreased admissions, shorter<br />
stays <strong>and</strong> lower costs. Chest. 1999;116:238S-239S.<br />
421
37. Wachter, RM. An introduction to the hospitalist model. Ann Intern Med. 1999; 134:338-<br />
342.<br />
38. Thompson DR, Clemmer TP, Applefeld JJ, Crippen DW, Wedel SK. Regionalization <strong>of</strong><br />
<strong>Critical</strong> Care Medicine. Crit Care Med. 1995;23:1154-1155.<br />
38. Pronovost PJ, Jenckes MW, Dorman T, Garrett E, Breslow MJ, Rosenfeld BA, et al.<br />
Organizational characteristics <strong>of</strong> intensive care units related to outcomes <strong>of</strong> abdominal<br />
aortic surgery. JAMA. 1999;281:1310-1317.<br />
39. Brown J, Sullivan G. Effect on ICU Mortality <strong>of</strong> a Full-time <strong>Critical</strong> Care Specialist. Chest.<br />
1989;96:127-129.<br />
40. Manthous CA, Amoateng-Adjepong Y, al-Kharrat T, Jacob B, Alnuaimat HM, Chatila W,<br />
et al. Effects <strong>of</strong> a medical intensivist on patient care in a community teaching hospital.<br />
Mayo Clin Proc. 1997;72:391-399.<br />
422
Chapter 39. Nurse Staffing, Models <strong>of</strong> Care Delivery, <strong>and</strong> Interventions<br />
Jean Ann Seago, PhD, RN<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Nursing<br />
Background<br />
Unlike the work <strong>of</strong> physicians, the work <strong>of</strong> registered nurses (RNs) in hospitals is rarely<br />
organized around disease-specific populations. Rather, patients are generally grouped by age<br />
<strong>and</strong>/or intensity <strong>of</strong> nursing care (eg, pediatrics or intensive care). Adult patients who require the<br />
least amount <strong>of</strong> nursing care (the largest proportion <strong>of</strong> hospitalized patients), may be separated<br />
into medical or surgical units but may also be combined on one unit. Because the work <strong>of</strong> RNs<br />
<strong>and</strong> other nurses is organized differently than the work <strong>of</strong> physicians, this chapter explores the<br />
literature related to nursing structure <strong>and</strong> process variables that may affect outcomes that relate<br />
to patient safety.<br />
Investigations <strong>of</strong> patient outcomes in relationship to nurses <strong>and</strong> their pr<strong>of</strong>essional<br />
responsibilities in hospitals commonly involve structural measures <strong>of</strong> care 1-4<br />
including numbers<br />
<strong>of</strong> nurses, number <strong>of</strong> nurse hours, percentage or ratios <strong>of</strong> nurses to patients, organization <strong>of</strong><br />
nursing care delivery or organizational culture, nurse workload, nurse stress, or qualification <strong>of</strong><br />
nurses. Less commonly, studies involve intervention or process measures <strong>of</strong> care including<br />
studies based on the science <strong>of</strong> nursing <strong>and</strong> others using nurses as the intervention. 1-5<br />
The use <strong>of</strong><br />
structural variables rather than process measures to study the impact <strong>of</strong> nursing activities reflects<br />
the greater availability <strong>of</strong> data relating to the former (<strong>of</strong>ten obtainable from administrative<br />
sources) compared with the latter (typically requiring chart review <strong>of</strong> direct observation). A<br />
number <strong>of</strong> structural measures have received considerable attention, specifically measures <strong>of</strong><br />
staffing levels in the face <strong>of</strong> major cost cutting <strong>and</strong> other changes in health care over the past 15-<br />
20 years. In 1996, the Institute <strong>of</strong> Medicine 6<br />
reported that there were insufficient data to draw<br />
conclusions about the relationship between nurse staffing <strong>and</strong> inpatient outcomes. However later<br />
studies have revisited this issue, allowing us to review the literature relating patient outcomes to<br />
various measures <strong>of</strong> nurse staffing levels, such as full time equivalents (FTEs), skill mix<br />
(proportion <strong>of</strong> RN hours to total hours), or RN hours per patient day.<br />
This chapter does not address patient outcomes as they relate to various “patient<br />
classification systems” (PCSs), although the prevalence <strong>of</strong> the use <strong>of</strong> such systems deserves<br />
mention. PCSs predict nursing care requirements at the individual patient level in order to<br />
determine unit staffing, project budgets, define an objective measure for costing out nursing<br />
services, <strong>and</strong> to maintain quality st<strong>and</strong>ards. 8<br />
Although PCSs are used for multiple purposes, they<br />
are an inadequate tool for determining unit staffing on a daily or shift basis. 9-11<br />
In addition, there<br />
are numerous patient classification systems 12-14<br />
<strong>and</strong> most are specific to one hospital or one<br />
nursing unit. The validity <strong>and</strong> reliability <strong>of</strong> PCSs are inconsistent <strong>and</strong> the systems cannot be<br />
compared with each other. 8-10,15-28<br />
Thus, rather than reviewing studies that analyze various PCS<br />
scores to patient outcomes, we review studies addressing the question <strong>of</strong> whether or not “safe<br />
thresholds” exist for levels <strong>of</strong> nursing care.<br />
423
Practice Description<br />
The availability <strong>of</strong> nurses, the organization <strong>of</strong> nursing care, <strong>and</strong> the types <strong>of</strong> nursing<br />
interventions vary by institution. Structuring nurse staffing (eg, availability <strong>of</strong> nurses,<br />
organizational models <strong>of</strong> nursing care) <strong>and</strong> care interventions to meet “safe thresholds” could be<br />
considered a patient safety practice. However, no studies have evaluated thresholds explicitly.<br />
This chapter reviews the precursor evidence from observational studies about the strength <strong>of</strong> the<br />
relationship between nursing variables <strong>and</strong> patient outcomes, so that possible safe thresholds<br />
may be inferred. We assess evidence that relates patient outcomes to:<br />
1) specific numbers, proportions, or ratios <strong>of</strong> nurses to patients (nurse staffing);<br />
Nurse availability variables generally characterize the number <strong>of</strong> hours nurses<br />
spend with patients. Typically, the time is not measured for each patient, but<br />
rather averages are measured based on the census <strong>of</strong> nurses to patients at a<br />
particular point in time. There are several common ways <strong>of</strong> accounting for this<br />
nurse staffing <strong>and</strong> no st<strong>and</strong>ardized way to measure it (Table 39.1).<br />
2) specific organization <strong>of</strong> nursing care delivery, nursing models <strong>of</strong> care, or<br />
organizational culture; Organization <strong>of</strong> nursing care variables (Table 39.2)<br />
may also include various nursing care delivery models, nursing unit or<br />
hospital culture, or governance structures. An issue <strong>of</strong> governance that has<br />
been studied by Aiken 29<br />
<strong>and</strong> others 30<br />
includes how much autonomy a nurse<br />
has to make practice decisions, how much control she has over practice<br />
decisions, how much collaboration occurs between physicians <strong>and</strong> nurse in<br />
the organization, <strong>and</strong> communication patterns; <strong>and</strong><br />
3) specific nursing interventions; Although nursing interventions are frequently<br />
studied in outpatient setting, 31,32-39<br />
perhaps because these venues provide<br />
nurses more flexibility to make independent decisions, 40-42<br />
studies in the<br />
inpatient setting have included measures <strong>of</strong> education, training, or retraining<br />
<strong>of</strong> nurses, providing audit data to nurses, <strong>and</strong> capturing nurse assessment <strong>of</strong><br />
patient outcomes.<br />
The varieties <strong>of</strong> intevention studies require some comment. Education interventions are<br />
popular in nursing research because they involve less risk than interventions that directly involve<br />
patients <strong>and</strong> are more readily approved by hospitals <strong>and</strong> physicians. 43-51<br />
Unfortunately, some<br />
investigators have made the assumption (which led to the failure to measure clinical outcomes)<br />
that increasing nursing knowledge or changing a practice, such as h<strong>and</strong>washing, automatically<br />
improves outcomes. 52,46,48,53<br />
Because a large part <strong>of</strong> a nurse’s job is assessment, investigators have used various<br />
nursing assessments as interventions, such as fall risk assessment, pressure ulcer risk assessment,<br />
or identification <strong>of</strong> patients at high risk for malnutrition, 55-60<br />
to reduce adverse events. In<br />
multidisciplinary protocols, the nursing activity is <strong>of</strong>ten assessment, rather than a nursing process<br />
or procedure. 49<br />
Other process-oriented interventions that lack sufficiently rigorous data to evaluate here,<br />
include specialty nurses, 61,62-65<br />
<strong>and</strong> interventions based on nursing science in the realm <strong>of</strong> nurse<br />
decision making in acute care hospitals (eg, mouth care to reduce mucositis, nonpharmacutical<br />
424
interventions to reduce pain, nausea <strong>and</strong> vomiting, increase sleep, <strong>and</strong> improve wound<br />
healing). 31,66-73<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The target safety problems are patient adverse events such as mortality <strong>and</strong> morbidity.<br />
The challenge is to create an optimum practice environment so that nurses can ideally reduce<br />
safety problems.<br />
Commonly studied adverse hospital events such as falls (Chapter 26), medication errors<br />
(Part III, Section A), <strong>and</strong> pressure ulcers (Chapter 27), are <strong>of</strong>ten used as outcome indicators for<br />
nursing practice. 83-90<br />
Less commonly studied are issues related to improving basic symptom<br />
management (eg, symptoms related to poor sleep, nutrition, or physical activity, or anxiety, pain,<br />
distress <strong>and</strong> discomfort caused by symptoms, or distress caused by diagnostic tests). In the last<br />
decade there has been increasing public <strong>and</strong> legislative pressure to improve hospital<br />
environments <strong>and</strong> address some <strong>of</strong> the heret<strong>of</strong>ore ignored issues. 91-93<br />
Opportunities for Impact<br />
Unfortunately, there is no definitive evidence as to specific thresholds for RN or total<br />
nursing staff hours per patient day, or nursing skill mix for various patient populations or nursing<br />
unit types. The lack <strong>of</strong> empirical evidence has been problematic for politicians, the public <strong>and</strong><br />
the nursing community. Because decisions about nurse staffing do not have a scientific basis <strong>and</strong><br />
are instead based on economics <strong>and</strong> anecdotes, nurse executives <strong>and</strong> managers are frequently at<br />
odds with staff nurses; especially those represented by labor unions, over staffing. Nurse<br />
executives are charged with providing safe patient care at a responsible cost. The need to<br />
constrain budgets by reducing nursing hours is in conflict with the needs <strong>of</strong> the unions <strong>and</strong>, some<br />
allege, in conflict with the needs <strong>of</strong> patients.<br />
Based in part on some limited data, New York <strong>and</strong> Massachusetts have passed legislation<br />
requiring formulae to be developed that ensure safe patient care. 95,96<br />
New Jersey has regulations<br />
which state that licensed nurses shall provide at least 65% <strong>of</strong> the direct care hours <strong>and</strong> requires<br />
an acuity system for patient classification. 97<br />
California Assembly Bill 394 directs the California<br />
Department <strong>of</strong> Health Services to establish nurse-to-patient staffing ratios for acute care<br />
hospitals by January 1, 2002. Sixteen states other than California have nurse staffing legislation<br />
on the calendar but have not implemented ratios. 94<br />
Staffing <strong>and</strong> ratios are items for collective bargaining <strong>and</strong> contract negotiations in some<br />
areas. 98-104<br />
Registering complains about “unsafe staffing” may be the nurses’ only recourse<br />
unless there is a negotiated agreement between the union <strong>and</strong> the hospital.<br />
Current utilization <strong>of</strong> practices using nursing interventions to make an impact on adverse<br />
hospital events is most likely limited due to uncertainty about effectiveness <strong>of</strong> specific<br />
interventions. Resources necessary for conducting systematic studies <strong>of</strong> nursing care provided in<br />
hospitals <strong>and</strong> then implementing the practices found to be helpful are scarce. 105-109<br />
Study Designs<br />
Searches <strong>of</strong> MEDLINE from 1990, CINHAL from 1966, documents published by the<br />
American Nurses Association, <strong>and</strong> the Cochrane Collaboration Library identified no r<strong>and</strong>omized<br />
clinical trials or non-r<strong>and</strong>omized controlled trials analyzing nurse staffing <strong>and</strong> adverse events.<br />
The study designs for nurse availability (Table 39.3) <strong>and</strong> organization <strong>of</strong> care (Table 39.4) are<br />
425
Level 2 or 3 designs. Mitchell et al 111<br />
references several r<strong>and</strong>omized trials in her review article.<br />
However, the articles mentioned used advanced practice nurses such as clinical nurse specialists,<br />
or home care visits as the intervention. 62,112,113<br />
The study by Jorgensen et al 114<br />
was set in a<br />
hospital but the comparison was between a specialty stroke unit <strong>and</strong> a regular care unit. The<br />
difference was between the different organization <strong>of</strong> stroke treatment, not nurse skill mix. The<br />
studies abstracted are observational studies that are case control, cohort, before-after, or health<br />
services research using data from large public databases.<br />
The study designs for nurse interventions (Table 39.5) vary from Level 1 to 3. Five<br />
studies use education <strong>of</strong> nurses as the intervention, <strong>and</strong> an additional 3 studies cover<br />
enhancements to education efforts (ie, providing data to nurses about adverse events in their<br />
units).<br />
Study Outcomes<br />
The studies <strong>of</strong> structural measures reported Level 1 or 2 outcomes, along with various<br />
other outcomes such as length <strong>of</strong> stay, patient satisfaction or nurse satisfaction. Most <strong>of</strong> the<br />
studies corrected for potential confounders <strong>and</strong> most adjusted outcomes based on patient acuity.<br />
The process measure studies vary between Level 2 <strong>and</strong> 3 outcomes. The studies also <strong>of</strong>ten<br />
included Level 4 outcomes, such as nurse knowledge, but these did not meet inclusion criteria.<br />
Most <strong>of</strong> the studies used adverse events such as falls, nosocomial infection, pain, phlebitis,<br />
medication errors or pressure ulcers as outcomes.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Nurse Staffing<br />
Table 39.4 summarizes the findings <strong>of</strong> studies exploring measures <strong>of</strong> nurse availability.<br />
When measured at the hospital level, there is mixed evidence that nurse staffing is related to 30day<br />
mortality. 30,83,115-118<br />
There is scarce but positive evidence that leaner nurse staffing is<br />
associated with unplanned hospital readmission <strong>and</strong> failure to rescue. 117,119-121<br />
There is strong<br />
evidence that leaner nurse staffing is associated with increased length <strong>of</strong> stay, nosocomial<br />
infection (urinary tract infection, postoperative infection, <strong>and</strong> pneumonia), <strong>and</strong> pressure<br />
ulcers. 122-125<br />
Results are conflicting as to whether richer nurse staffing has a positive effect on patient<br />
outcomes. Although 5 30,89,118,120,129<br />
<strong>of</strong> the 16 studies in Table 39.3 reported no association<br />
between richer nurse staffing <strong>and</strong> positive patient outcomes, the other 11 that report an<br />
association tend to be more recent, with larger samples <strong>and</strong> more sophisticated methods for<br />
accounting for confounders. These studies had various types <strong>and</strong> acuities <strong>of</strong> patients <strong>and</strong>, taken<br />
together, provide substantial evidence that richer nurse staffing is associated with better patient<br />
outcomes. Although the optimum range for acute care hospital nursing staffing is most likely<br />
within these ranges, none <strong>of</strong> the studies specifically identify the ratios or hours <strong>of</strong> care that<br />
produce the best outcomes for different groups <strong>of</strong> patients or different nursing units.<br />
Models <strong>of</strong> Nursing Care Delivery<br />
The 7 studies in Table 39.4 provide mixed evidence about the relationship between<br />
organization <strong>of</strong> nursing care <strong>and</strong> patient outcomes. Aiken et al 29<br />
found that hospitals with<br />
“magnet” characteristics have lower mortality in one study, but not in another, 115<br />
<strong>and</strong> Shortell et<br />
426
al 30<br />
also does not find an association in ICUs. Seago 79<br />
found a reduction in medication errors<br />
after a change to patient-focused care <strong>and</strong> Grillo-Peck et al 130<br />
found a reduction in falls after a<br />
change to a RN-UAP (unlicensed assistive personnel) partner model was introduced. The 2<br />
review articles 111,131<br />
reported mixed results about whether nursing models, nurse surveillance or<br />
work environment is associated with patient outcomes. Thus, the evidence is insufficient to<br />
direct practice.<br />
Nursing Interventions<br />
Table 39.5 provides details about studies using nurse interventions. The first 3 studies<br />
provide support for the idea that added education <strong>of</strong> nurses reduces infection <strong>and</strong><br />
thrombophlebitis. The subsequent 2 studies, however, found no difference in bloodstream<br />
infection or medication error before <strong>and</strong> after added education. The overall evidence indicates<br />
that using education as the sole intervention does not always change patient outcomes.<br />
Educational interventions were related to changes in nurse practices <strong>and</strong>, in some studies, also<br />
related to decreasing adverse events. 44,47,54<br />
However adding another intervention such as providing<br />
feedback data or benchmarking results, was more likely to be associated with improved patient<br />
outcomes, 55-57<br />
including decreased infection rates, pressure ulcer rates, <strong>and</strong> fall rates. 55-57<br />
Potential for Harm<br />
The potential for harm <strong>of</strong> patients associated with structural interventions such as too few<br />
nurses has been documented. 83-85,124,125<br />
Studies involving process interventions such as using<br />
education <strong>of</strong> nurses, providing data to nurses, <strong>and</strong> interventions based on nursing science, seem<br />
to have a low probability <strong>of</strong> harm, but that is as yet unknown.<br />
Costs <strong>and</strong> Implementation<br />
Few <strong>of</strong> the abstracted studies mentioned cost, although several measured length <strong>of</strong> stay as<br />
an outcome variable. Pratt et al 63<br />
found no difference in quality <strong>of</strong> care measures using a 100%<br />
RN skill mix <strong>and</strong> an 80% RN skill mix in 2 wards in one hospital in the United Kingdom. The<br />
cost was less with the 80% skill mix but the nurses who worked with less experienced staff<br />
reported an increase in workload <strong>and</strong> increase in stress. California is faced with impending<br />
legislated minimum nurse staffing ratios in the acute care hospitals. Based on early studies, 149<br />
at<br />
least 40% <strong>of</strong> California hospitals may see a negative financial effect because <strong>of</strong> the need to<br />
increase staffing. Additionally, based on a number <strong>of</strong> predictions, 150,151<br />
there is now, <strong>and</strong> there<br />
will continue to be, a significant shortage <strong>of</strong> registered nurses in the US. Thus, implementing any<br />
increase in RN staffing may be very difficult.<br />
One investigator who provided data to nurses as the intervention related to urinary<br />
catheter infection reported an estimated cost savings <strong>of</strong> $403,000. 55<br />
Another investigator who<br />
also provided data to nurses related to nosocomial pressure ulcer rates estimated implementation<br />
costs but not cost saving. 57<br />
The investigator who studied adding an IV team (specialty nurses)<br />
reported a savings <strong>of</strong> $53,000/saved life <strong>and</strong> $14,000/bloodstream infection. Using clean rather<br />
than sterile dressings on open postoperative wounds saved $9.59/dressing with no change in rate<br />
<strong>of</strong> wound healing. Based on these studies, it is likely that some nursing interventions can save<br />
costs.<br />
427
Comment<br />
The studies evaluated in this review include only medical, surgical <strong>and</strong> ICU nursing<br />
units. Other data from more specialized units, the outpatient setting, <strong>and</strong> those pertaining to<br />
subsets <strong>of</strong> patients tend to mirror the findings <strong>of</strong> the evidence evaluation, <strong>and</strong> are cited in this<br />
section alongside those abstracted <strong>and</strong> presented in the evidence tables.<br />
The relationship <strong>of</strong> hospital environment to patient outcomes is still being debated.<br />
However, evidence using hospital-level data indicates increasing the percentage <strong>of</strong> RNs in the<br />
skill mix, increasing RN FTEs or hours per patient day or average daily census is associated with<br />
decreased risk-adjusted mortality. 116,131,152,153<br />
Other studies, also aggregating data to the hospital<br />
level, found that increasing RN hours per patient day is associated with decreased nosocomial<br />
infection rates, 121,154<br />
decreased urinary tract infections, thrombosis <strong>and</strong> pulmonary complications<br />
in surgical patients, 124<br />
decreased pressure ulcers, pneumonia, postoperative infection <strong>and</strong> urinary<br />
tract infection. 122,125<br />
Hunt 117<br />
found that decreasing ratios were related to increasing readmission<br />
rates but were not related to mortality rates.<br />
The cost <strong>of</strong> primary data collection has limited the number <strong>of</strong> studies using data<br />
aggregated to the individual nursing unit. There is some evidence that decreased nurse-to-patient<br />
ratios in the ICU was associated with an increase in blood stream infections associated with<br />
central venous catheter, 126<br />
while an increase in agency nurses was related to other negative<br />
patient outcomes. 156<br />
A study in the NICU setting found understaffing <strong>and</strong> overcrowding <strong>of</strong><br />
patients led to an outbreak <strong>of</strong> Enterobactor cloacae. 155<br />
In 42 ICUs Shortell et al. found that low<br />
nurse turnover was related to shorter length <strong>of</strong> stay 30<br />
; in 65 units an increase in nurse<br />
absenteeism was related to an increase in urinary tract infection <strong>and</strong> other patient infections but<br />
not to other adverse events. 157<br />
Amaravadi et al 158<br />
found that night nurse-to-patient ratio in ICUs<br />
in 9 hospitals for a select group <strong>of</strong> patients who had undergone esophagectomy was not<br />
associated with mortality but was associated with a 39% increase in length <strong>of</strong> stay <strong>and</strong> higher<br />
pneumonia rates, reintubation rates, <strong>and</strong> septicimia rates. As noted previously, Blegan et al found<br />
that as the percentage <strong>of</strong> RNs per total staff (skill mix) increased there was a decrease in<br />
medication errors, decubitus ulcers, <strong>and</strong> patient complaints up to a skill mix <strong>of</strong> 85-87% RNs. 83,84<br />
In several studies, increasing skill mix was associated with decreasing falls, length <strong>of</strong><br />
stay, postoperative complications, nosocomial pneumonia, pressure ulcer rates, urinary tract<br />
infection, <strong>and</strong> postoperative infection. 122-125,130<br />
Several studies with varying sample sizes have<br />
found skill mix to be unrelated to mortality. 111,118,159,160<br />
Others have found skill mix to be<br />
unrelated to treatment problems, postoperative complications, unexpected death rates, or<br />
unstable condition at discharge 129<br />
<strong>and</strong> found no relationship between skill mix or nursing hours<br />
per patient day <strong>and</strong> medication errors, falls, patient injuries, <strong>and</strong> treatment errors. 161<br />
In an early<br />
study <strong>of</strong> primary (all RN) <strong>and</strong> team (skill mix) nursing care delivery models, there was no<br />
relationship between percent <strong>of</strong> RNs <strong>and</strong> quality <strong>of</strong> care as measured by nurse report 162<br />
<strong>and</strong> in 23<br />
hospitals in the Netherl<strong>and</strong>s, there was no relationship between RN-to-patient ratio <strong>and</strong> incidence<br />
<strong>of</strong> falls. 89<br />
Although mixed, the overall evidence seems to indicate that proportion <strong>of</strong> RN hours per<br />
total hours <strong>and</strong> richer RN-to-patient ratios likely do not affect 30-day mortality, may be<br />
428
associated with in-hospital mortality, <strong>and</strong> are probably associated with adverse events such as<br />
postoperative complications, nosocomial infection, medication errors, falls, <strong>and</strong> decubitus ulcers.<br />
Based on recent work, nurse staffing was examined in “best practices” hospitals. This<br />
included hospitals recognized by the American Nurses Association’s Magnet Hospital program,<br />
those commended by the Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations<br />
(JCAHO), those listed in USA Today’s Top 100 Hospitals, those listed in US News <strong>and</strong> World<br />
Report’s set <strong>of</strong> high-quality hospitals, those noted for having better than expected mortality for<br />
heart attacks <strong>and</strong> newborn readmission rates by the Pacific Business Group on Health (PBGH),<br />
<strong>and</strong> those recognized by the Bay Area Consumer Checkbook for high quality. There is<br />
significant variation in nurse staffing among these best practices hospitals. The staffing data for<br />
best practices hospitals do not consistently demonstrate that hospitals rated highly for quality <strong>of</strong><br />
patient care have uniformly richer staffing than do other hospitals. 74<br />
Because units within<br />
hospitals vary widely in nurse staffing <strong>and</strong> outcomes, results from data aggregated to the hospital<br />
level are difficult to interpret.<br />
At present the literature is insufficient to make a reasoned judgment about organization<br />
<strong>of</strong> the work environment <strong>of</strong> nurses. Further work is needed in the area <strong>of</strong> nurse interventions. If<br />
there truly is to be an emphasis on reducing adverse events in hospitals <strong>and</strong> creating hospital<br />
environments that promote health <strong>and</strong> healing, resources for research related to nurses <strong>and</strong><br />
nursing interventions must be found.<br />
429
Table 39.1. Measures <strong>of</strong> nurse staffing<br />
Nurse Staffing<br />
Measure<br />
Definition<br />
Nurse to patient ratio Number <strong>of</strong> patients cared for by one nurse typically specified by<br />
job category (RN, Licensed Vocational or Practical Nurse-LVN or<br />
LPN); this varies by shift <strong>and</strong> nursing unit; some researchers use<br />
this term to mean nurse hours per inpatient day<br />
Total nursing staff or<br />
hours per patient day<br />
RN or LVN FTEs per<br />
patient day<br />
Nursing skill (or staff)<br />
mix<br />
Table 39.2. Models <strong>of</strong> nursing care delivery<br />
Nursing Care Delivery<br />
Models<br />
All staff or all hours <strong>of</strong> care including RN, LVN, aides counted per<br />
patient day (a patient day is the number <strong>of</strong> days any one patient<br />
stays in the hospital, ie, one patient staying 10 days would be 10<br />
patient days)<br />
RN or LVN full time equivalents per patient day (an FTE is 2080<br />
hours per year <strong>and</strong> can be composed <strong>of</strong> multiple part-time or one<br />
full-time individual)<br />
The proportion or percentage <strong>of</strong> hours <strong>of</strong> care provided by one<br />
category <strong>of</strong> caregiver divided by the total hours <strong>of</strong> care (A 60% RN<br />
skill mix indicates that RNs provide 60% <strong>of</strong> the total hours <strong>of</strong> care)<br />
Definition<br />
<strong>Patient</strong> Focused Care A model popularized in the 1990s that used RNs as care managers<br />
<strong>and</strong> unlicensed assistive personnel (UAP) in exp<strong>and</strong>ed roles such as<br />
drawing blood, performing EKGs, <strong>and</strong> performing certain<br />
assessment activities<br />
Primary or Total<br />
Nursing Care<br />
Team or Functional<br />
Nursing Care<br />
Magnet Hospital<br />
Environment/Shared<br />
governance<br />
A model that generally uses an all-RN staff to provide all direct<br />
care <strong>and</strong> allows the RN to care for the same patient throughout the<br />
patient’s stay; UAPs are not used <strong>and</strong> unlicensed staff do not<br />
provide patient care<br />
A model using the RN as a team leader <strong>and</strong> LVNs/UAPs to<br />
perform activities such as bathing, feeding, <strong>and</strong> other duties<br />
common to nurse aides <strong>and</strong> orderlies; it can also divide the work by<br />
function such as “medication nurse” or “treatment nurse”<br />
Characterized as “good places for nurses to work” <strong>and</strong> includes a<br />
high degree <strong>of</strong> RN autonomy, MD-RN collaboration, <strong>and</strong> RN<br />
control <strong>of</strong> practice; allows for shared decision making by RNs <strong>and</strong><br />
managers<br />
430
Table 39.3 Structural measures: availability <strong>of</strong> nurses <strong>and</strong> patient outcomes (First 11<br />
studies showed positive associations; final 5 studies detected no significant effect)<br />
Study Setting Study<br />
Design,<br />
Outcomes<br />
1. Data were collected form<br />
1,205 consecutively admitted<br />
patients in 40 units in 20 acute<br />
care hospitals <strong>and</strong> on 820 nurses<br />
in the US 115<br />
2. All patients who developed a<br />
central venous catheter<br />
bloodstream infection during an<br />
infection outbreak period<br />
(January 1992 through<br />
September 1993) <strong>and</strong> r<strong>and</strong>omly<br />
selected controls. Cohort study:<br />
all SICU patients during the<br />
study period (January 1991<br />
through September 1993) 126<br />
3. 39 nursing units in 11<br />
hospitals for 10 quarters <strong>of</strong> data<br />
between July, 1993 <strong>and</strong><br />
December, 1995 in the US 84<br />
Level 3,<br />
Level 1&3<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1&2<br />
Availability <strong>of</strong><br />
Nurses<br />
0.8 mean nurse/<br />
patient day with a<br />
range <strong>of</strong> 0.5-1.5<br />
nurses/patient day<br />
1.2 patient/nurse<br />
<strong>and</strong> 20 nursing<br />
hours per patient<br />
day (HPPD)<br />
1.5 patient/nurse<br />
<strong>and</strong> 16 nursing<br />
HPPD<br />
2 patient/nurse <strong>and</strong><br />
12 nursing HPPD<br />
Proportion <strong>of</strong><br />
direct care RN<br />
hours; total direct<br />
care hours;<br />
Up to 87.5% RN<br />
skill mix<br />
431<br />
Effect Size (coefficient, mean<br />
differences, OR)<br />
This measure was significantly<br />
associated with 30-day mortality<br />
(OR .46, 95% CI: 0.22-0.98). An<br />
additional nurse per patient day<br />
reduces the odds <strong>of</strong> dying by onehalf.<br />
There was a significant<br />
relationship between nurse to<br />
patient ratios <strong>and</strong> nursing hours<br />
<strong>and</strong> central venous catheter<br />
bloodstream infection in the SICU.<br />
For 1.2 patients/nurse <strong>and</strong> 20<br />
HPPD the adjusted odds ratio was<br />
3.95 (95% CI: 1.07-14.54), 1.5<br />
patients/nurse <strong>and</strong> 16 nursing<br />
HPPD, 15.6 (95% CI: 1.15-211.4),<br />
<strong>and</strong> for 2 patients/nurse <strong>and</strong> 12<br />
HPPD, 61.5 (95% CI:1.23-3074).<br />
With patient acuity controlled,<br />
direct care RN proportion <strong>of</strong> hours<br />
was inversely associated with<br />
medication errors (-0.525 p
4. 42 inpatient units in one 880bed<br />
hospital in the US 83<br />
5. Data from hospital cost<br />
disclosure reports <strong>and</strong> patient<br />
discharge abstracts from acute<br />
care hospitals in California <strong>and</strong><br />
New York for fiscal years 1992<br />
<strong>and</strong> 1994 125<br />
6. Data from hospital cost<br />
disclosure reports, patient<br />
discharge abstracts <strong>and</strong> Medicare<br />
data from acute care hospitals in<br />
Arizona, California, Florida,<br />
Massachusetts, New York, <strong>and</strong><br />
Virginia for 1996 123<br />
7. Data from hospital cost<br />
disclosure reports, patient<br />
discharge abstracts from acute<br />
care hospitals in California,<br />
Massachusetts, <strong>and</strong> New York<br />
for 1992 <strong>and</strong> 1994 122<br />
Level 3,<br />
Level 1&2<br />
Level 3,<br />
Level 1&2<br />
Level 3,<br />
Level 1&2<br />
Level 3,<br />
Level 1&2<br />
8.63 mean total<br />
hours <strong>of</strong> care;<br />
69% RN skill mix;<br />
up to 85% skill<br />
mix<br />
7.56-8.43 mean<br />
total hours <strong>of</strong><br />
care/nursing<br />
intensity weight<br />
(NIW); 67.7% to<br />
70.5% RN skill<br />
mix<br />
5.76 mean<br />
licensed hours <strong>of</strong><br />
care/ 83.3% RN<br />
skill mix<br />
7.67-8.43 mean<br />
total hours <strong>of</strong> care;<br />
67.7-70.5% skill<br />
mix<br />
432<br />
With patient acuity controlled,<br />
direct care RN proportion <strong>of</strong> hours<br />
was inversely associated with<br />
medication errors/doses (-<br />
0.576, p
8. Data from HCFA Medicare<br />
Hospital Mortality Information<br />
1986 <strong>and</strong> the American Hospital<br />
Association 1986 annual survey<br />
<strong>of</strong> hospitals 116<br />
9. Data from the American<br />
Hospital Association 1986<br />
annual survey <strong>of</strong> hospitals <strong>and</strong><br />
medical record reviews from<br />
July 1987 to June 1988 in 6 large<br />
PPOs 128<br />
10. Data from the American<br />
Hospital Association Annual<br />
Survey <strong>of</strong> Hospitals for 1993 <strong>and</strong><br />
the Nationwide Inpatient Sample<br />
from the Agency for Health Care<br />
Policy <strong>and</strong> Research for 1993<br />
(HCUP-3) 124<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 3<br />
Level 3,<br />
Level 1<br />
0.9 mean<br />
RN/ADC (average<br />
daily census); 60%<br />
skill mix<br />
52.2 (Texas)-<br />
67.6% (California)<br />
skill mix<br />
67.8% mean skill<br />
mix<br />
433<br />
Controlling for hospital<br />
characteristics, number <strong>of</strong><br />
RNs/ADC was not significantly<br />
related to adjusted 30-day<br />
mortality rate but proportion <strong>of</strong><br />
RNs/all nursing staff was<br />
significantly related to adjusted<br />
30-day mortality rate (adjusted<br />
difference between lower <strong>and</strong><br />
upper fourth <strong>of</strong> hospitals -2.5, 95%<br />
CI: -4.0 to -0.9)<br />
Controlling for hospital<br />
characteristics, number <strong>of</strong><br />
RNs/ADC was not significantly<br />
related to problem rate but<br />
proportion <strong>of</strong> RNs/all nursing staff<br />
was significantly related to lower<br />
problem rates (California lower<br />
rates 3.58, upper rates 2.30<br />
p
11. Data were collected form<br />
March 1 to June 7, 1986 <strong>and</strong><br />
included 497 patients 127<br />
12. 390 patients admitted within<br />
1 week after stroke onset in 9<br />
acute care hospitals in The<br />
Netherl<strong>and</strong>s. Surviving patients<br />
were interviewed 6 months poststroke<br />
<strong>and</strong> asked about falls. Fall<br />
<strong>and</strong> other patient data were<br />
collected from medical records.<br />
Ward characteristics were<br />
provided by senior nurses. There<br />
is complete data on 349<br />
patients 89<br />
13. 17,440 patients across 42<br />
ICUs in the US 30<br />
14. Data were collected from<br />
April, 1994-March, 1995 from<br />
23 trusts (groups <strong>of</strong> hospitals) in<br />
Scotl<strong>and</strong> 117<br />
15. Data were collected form the<br />
American Hospital Association<br />
Annual Survey <strong>of</strong> Hospitals in<br />
1989-1991, the observed <strong>and</strong><br />
predicted 30-day post-admission<br />
mortality for patients with a<br />
primary diagnosis <strong>of</strong> COPD<br />
from the HCFA Hospital<br />
Information Reports from 1989-<br />
1991 <strong>and</strong> the Medicare Case Mix<br />
Index 118<br />
16. Data from staffing <strong>and</strong><br />
accounting records <strong>of</strong> 60<br />
community hospitals across the<br />
US in 1985, hospital <strong>and</strong> nursing<br />
unit surveys, 1981 case mix<br />
indexes from the Federal<br />
Register, <strong>and</strong> the Health Area<br />
Resources File 129<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 1-3<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 3<br />
Adequate staffing The adequately staffed unit had<br />
fewer complications than the<br />
inadequately staffed unit.<br />
0.04 mean<br />
difference in nurse<br />
to patient ratios<br />
Mean .66<br />
patient/nurse with<br />
a range <strong>of</strong> 0.31-<br />
1.31<br />
Mean RN FTE<br />
was 1.21 per<br />
patient<br />
RN FTE/100<br />
adjusted<br />
admissions<br />
52% RN skill mix;<br />
33% LPN mean<br />
nursing HPPD was<br />
4.93<br />
434<br />
There was no statistical difference<br />
in falls between case <strong>and</strong> control<br />
groups in number <strong>of</strong> nurses or<br />
nurse ratios on any shift. Days<br />
(mean difference -0.06, CI: -<br />
0.51 to 0.39); Evening (mean<br />
difference -0.24, 95% CI: -0.97 to<br />
0.50); Nights (mean difference<br />
1.24, 95% CI: 0.28 to 2.20); All<br />
shifts (mean difference 0.04, 95%<br />
CI, -0.33 to 0.40).<br />
Neither nurse to patient ratio nor<br />
caregiver interaction was found to<br />
be significantly associated with<br />
risk-adjusted mortality.<br />
There was no association between<br />
RN FTE per occupied hospital bed<br />
<strong>and</strong> mortality<br />
There was no association between<br />
RN FTE/100 adjusted admissions<br />
<strong>and</strong> 30-day post-admission<br />
mortality for patients with a<br />
primary diagnosis <strong>of</strong> COPD<br />
None <strong>of</strong> the staffing variables <strong>of</strong><br />
interest were associated with<br />
medication errors, patient injuries,<br />
IV administration errors, or<br />
treatment errors.
Table 39.4 Structural variables: nursing organization models <strong>and</strong> patient outcomes<br />
Study Setting Study<br />
Design,<br />
Outcomes<br />
Data were collected from 39<br />
"magnet" hospitals, which are<br />
hospitals designated as good<br />
places for nurses to work, <strong>and</strong><br />
195 nonmagnet matched<br />
hospitals 29<br />
Data were collected form 1,205<br />
consecutively admitted patients<br />
in 40 units in 20 acute care<br />
hospitals <strong>and</strong> on 820 nurses in<br />
the US 115<br />
17,440 patients across 42 ICUs<br />
in the US 30<br />
Data were collected at 3 points in<br />
time; 6 month before the<br />
intervention, 6 months, <strong>and</strong> 12<br />
months after the introduction <strong>of</strong><br />
the new model <strong>and</strong> included the<br />
time between October 1996 to<br />
December 1997 79<br />
Data were collected 6 months<br />
before <strong>and</strong> 6 months after the<br />
introduction <strong>of</strong> the new model<br />
<strong>and</strong> included the time between<br />
January-June, 1992 <strong>and</strong> January-<br />
June, 1993 130<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 1&2<br />
Level 3,<br />
Level 1-3<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
Review article: Pierce, 1997131 Level 3A,<br />
Level 1&2<br />
Organization<br />
<strong>of</strong> Care/Models<br />
435<br />
Effect Size (coefficient, mean<br />
differences, OR)<br />
Magnet hospitals Magnet hospitals had a 4.6% lower<br />
adjusted Medicare mortality rates<br />
(p=0.026, 95% CI: 0.9-9.4 fewer<br />
deaths per 1,000)<br />
Magnet hospitals<br />
(nurse control<br />
over practice<br />
variable)<br />
Magnet hospitals<br />
(nurse unit<br />
culture captured<br />
in caregiver<br />
interaction<br />
variable)<br />
<strong>Patient</strong> Focused<br />
Care<br />
RN-UAP<br />
Partnership<br />
similar to <strong>Patient</strong><br />
Focused Care<br />
Nursing<br />
Environment<br />
Nurse control over practice was<br />
not significantly associated with<br />
any clinical outcomes, but was<br />
significantly associated with<br />
patient satisfaction (coefficient<br />
0.56 (95% CI: 0.16-97)<br />
Caregiver interaction was not<br />
significantly associated with<br />
clinical outcomes, but was<br />
significantly associated with lower<br />
risk-adjusted length <strong>of</strong> stay (-0.16,<br />
p
Review article: MEDLINE from<br />
1966-1996, CINAHL from 1982-<br />
1996, Exp<strong>and</strong>ed Academic Index<br />
from 1989-1996, search by<br />
author for investigators known to<br />
be working in the field, manual<br />
searches <strong>of</strong> the bibliographies <strong>of</strong><br />
review articles <strong>and</strong> monographs<br />
(Mitchell) 111<br />
Level 3A,<br />
Level 1&2<br />
Nursing<br />
Environment<br />
436<br />
Mixed results in studies about<br />
whether nursing surveillance,<br />
quality <strong>of</strong> working environment,<br />
<strong>and</strong> quality <strong>of</strong> interaction with<br />
other pr<strong>of</strong>essionals predict<br />
hospitals with lower mortality.<br />
With more sophisticated risk<br />
adjustment, evidence suggests that<br />
mortality <strong>and</strong> complications are<br />
related more to patient variables<br />
<strong>and</strong> adverse events may be more<br />
closely related to organizational<br />
characteristics.
Table 39.5 Process measures: nurse intervention <strong>and</strong> patient outcomes<br />
Study Setting Study<br />
Design,<br />
Outcomes<br />
Data were collected from 60<br />
hospitalized patients on 1 surgical<br />
service in a university hospital in<br />
Turkey between September 1996<br />
<strong>and</strong> September 1997 44<br />
2 surgical <strong>and</strong> 2 medical wards in<br />
one hospital in Sweden were<br />
r<strong>and</strong>omly assigned to either a<br />
control or experimental group. 18<br />
nurses on the experimental wards<br />
<strong>and</strong> 18 nurse on the control wards;<br />
90 patients on the experimental<br />
wards <strong>and</strong> 39 patients on the control<br />
wards; 112 Peripheral IVs on the<br />
experimental wards <strong>and</strong> 60 PIVs on<br />
the control wards 47<br />
One hospital in Spain; all<br />
nosocomial infection data between<br />
March 1982 <strong>and</strong> December 1990 54<br />
One university hospital in<br />
Washington, DC; all adult patients<br />
with bloodstream Infections<br />
between July 1984 <strong>and</strong> February<br />
1994 (n=432) 45<br />
One general hospital in Illinois; all<br />
omitted <strong>and</strong> wrong dose medication<br />
errors between October 1992 <strong>and</strong><br />
March 1993 43<br />
Level 2,<br />
Level 2&3<br />
Level 1,<br />
Level 2&3<br />
Level 3,<br />
Level 1<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
437<br />
Intervention Effect Size (coefficient, mean<br />
differences, OR)<br />
Added<br />
education to<br />
intervention<br />
group<br />
Added<br />
education to<br />
intervention<br />
group<br />
Added<br />
education to<br />
intervention<br />
group<br />
Added<br />
education<br />
Added<br />
education<br />
Positive colonization <strong>of</strong> catheter<br />
hub was 68.6% in the control<br />
group <strong>and</strong> 25% in the intervention<br />
group (chi square=5.75, p
All urinary catheter-patient-days<br />
between January 1995 <strong>and</strong><br />
September 1996 in 1 VA hospital 55<br />
Stanford University Hospital; all<br />
pressure ulcers <strong>and</strong> nosocomial<br />
pressure ulcers during 1992 through<br />
1996 57<br />
8. Stanford University Hospital 52<br />
bed medical surgical unit; all falls<br />
between 1995 through 1996 56<br />
References<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Level 2<br />
438<br />
Provided<br />
infection rate<br />
data to nurses<br />
Provided<br />
nosocomial<br />
pressure rate<br />
data to nurses<br />
plus added<br />
education<br />
Provided fall<br />
rate data to<br />
nurses <strong>and</strong><br />
added<br />
education<br />
Pre-intervention there were<br />
32/1000 catheter-patient days<br />
(95% CI: 22.9-43.7); for the 5<br />
quarters post intervention, there<br />
was a significant decrease<br />
(p
10. Finnigan SA, Abel M, Dobler T, Hudon L, Terry B. Automated patient acuity. Linking<br />
nursing systems <strong>and</strong> quality measurement with patient outcomes. J Nurs Adm. 1993;23:62-<br />
71.<br />
11. Lawson KO, Formella NM, Smeltzer CH, Walters RM. Redefining the purpose <strong>of</strong> patient<br />
classification [published erratum appears in Nurs Econ 1993 Nov-Dec;11(6):382]. Nurs<br />
Econ. 1993;11:298-302.<br />
12. Cohen MM, O'Brien-Pallas LL, Copplestone C, Wall R, Porter J, Rose DK. Nursing<br />
workload associated with adverse events in the postanesthesia care unit. Anesthesiology.<br />
1999;91:1882-1890.<br />
13. Garfield M, Jeffrey R, Ridley S. An assessment <strong>of</strong> the staffing level required for a highdependency<br />
unit. Anaesthesia. 2000;55:137-143.<br />
14. Moreno R, Reis Mir<strong>and</strong>a D. Nursing staff in intensive care in Europe: the mismatch<br />
between planning <strong>and</strong> practice. Chest. 1998;113:752-758.<br />
15. DeGroot HA. <strong>Patient</strong> classification systems <strong>and</strong> staffing. Part 2, Practice <strong>and</strong> process. J<br />
Nurs Adm. 1994;24:17-23.<br />
16. DeGroot HA. <strong>Patient</strong> Classification System Evaluation: Part 2, System Selection <strong>and</strong><br />
Implemention. J <strong>of</strong> Nurs Admin. 1989;19:24-30.<br />
17. DeGroot HA. <strong>Patient</strong> classification systems <strong>and</strong> staffing. Part 1, Problems <strong>and</strong> promise. J<br />
Nurs Adm. 1994;24:43-51.<br />
18. Davidhizar R, Mallow GE, Bechtel GA, Giger JN. A patient classification system for the<br />
chronic psychiatric patient. Aust N Z J Ment Health Nurs. 1998;7:126-133.<br />
19. Detwiler C, Clark MJ. Acuity classification in the urgent care setting. J Nurs Adm.<br />
1995;25:53-61.<br />
20. Dunbar LJ, Diehl BC. Developing a patient classification system for the pediatric<br />
rehabilitation setting. Rehabil Nurs. 1995;20:328-332.<br />
21. Freund L, Burrows-Hudson S, Preisig P. Development <strong>of</strong> a patient classification system for<br />
chronic hemodialysis patients. Am J Kidney Dis. 1998;31:818-829.<br />
22. Giovannetti P, Johnson JM. A new generation <strong>of</strong> patient classification system. JONA.<br />
1990;20:33-40.<br />
23. Godin M. A patient classification system for the hemodialysis setting. Nurs Manage.<br />
1995;26:66-67.<br />
24. Lovett RB, Reardon MB, Gordon BK, McMillan S. Validity <strong>and</strong> reliability <strong>of</strong> medical <strong>and</strong><br />
surgical oncology patient acuity tools. Oncol Nurs Forum. 1994;21:1709-1717.<br />
25. Lovett RB, Wagner L, McMillan S. Validity <strong>and</strong> reliability <strong>of</strong> a pediatric hematology<br />
oncology patient acuity tool. J Pediatr Oncol Nurs. 1991;8:122-130.<br />
26. O'Brien-Pallas L, Irvine D, Peereboom E, Murray M. Measuring nursing workload:<br />
underst<strong>and</strong>ing the variability. Nursing Economics. 1997;15:171-182.<br />
27. O'Brien-Pallas LL, Cockerill RW. Satisfaction with nursing workload systems: report <strong>of</strong> a<br />
survey <strong>of</strong> Canadian hospitals. Part B. Can J Nurs Adm. 1990;3:23-26.<br />
28. Walts LM, Kapadia AS. <strong>Patient</strong> classification system: an optimization approach. Health<br />
Care Manage Rev. 1996;21:75-82.<br />
29. Aiken LH, Smith HL, Lake ET. Lower Medicare mortality among a set <strong>of</strong> hospitals known<br />
for good nursing care. Med Care. 1994;32:771-787.<br />
30. Shortell SM, Zimmerman JE, Rousseau DM, et al. The performance <strong>of</strong> intensive care units:<br />
does good management make a difference? <strong>Medical</strong> Care. 1994;32:508-525.<br />
439
31. Dodd MJ, Dibble SL, Miaskowski C, et al. R<strong>and</strong>omized clinical trial <strong>of</strong> the effectiveness <strong>of</strong><br />
3 commonly used mouthwashes to treat chemotherapy-induced mucositis. Oral Surgery,<br />
Oral Medicine, Oral Pathology, Oral Radiology <strong>and</strong> Endodontics. 2000;90:39-47.<br />
32. Robertson R, Samuelson C. Should nurse patient ratios be legislated? - Pros <strong>and</strong> cons -.<br />
Georgia Nursing. 1996;56:2.<br />
33. Shoemake A. Solutions to the home healthcare nursing shortage. Home Healthc Nurse.<br />
1994;12:35-39.<br />
34. Davis MA, Freeman JW. Excess dem<strong>and</strong> <strong>and</strong> cost relationships among Kentucky nursing<br />
homes. Health Care Financ Rev. 1994;15:137-152.<br />
35. Carr EC. Talking on the telephone with people who have experienced pain in hospital:<br />
clinical audit or research? J Adv Nurs. 1999;29:194-200.<br />
36. Waterman H, Waterman C. Trends in ophthalmology services, nursing skill-mix <strong>and</strong><br />
education: 2nd national survey. J Adv Nurs. 1999;30:942-949.<br />
37. Miskella C, Avis M. Care <strong>of</strong> the dying person in the nursing home: exploring the care<br />
assistants' contribution including commentary by Froggatt K <strong>and</strong> Sale DNT. European<br />
Journal <strong>of</strong> Oncology Nursing. 1998;2:80-88.<br />
38. Eliopoulos C. Nurse staffing in long-term care facilities: the case against a high ratio <strong>of</strong><br />
RNs. J Nurs Adm. 1983;13:29-31.<br />
39. Gagnon AJ, Edgar L, Kramer MS, Papageorgiou A, Waghorn K, Klein MC. A r<strong>and</strong>omized<br />
trial <strong>of</strong> a program <strong>of</strong> early postpartum discharge with nurse visitation. Am J Obstet<br />
Gynecol. 1997;176:205-211.<br />
40. Cumbey DA, Alex<strong>and</strong>er JW. The relationship <strong>of</strong> job satisfaction with organizational<br />
variables in public health nursing. J Nurs Adm. 1998;28:39-46.<br />
41. Schmidt CE, Gillies DA, Biordi D, Child DA. Marketing the home healthcare agency. Do<br />
nurses <strong>and</strong> physicians agree? J Nurs Adm. 1990;20:9-17.<br />
42. Jansen PGM, Kerkstra A, Abu Saad HH, van der Zee J. Models <strong>of</strong> differentiated practice<br />
<strong>and</strong> specialization in community nursing: a review <strong>of</strong> the literature. J Adv Nurs.<br />
1996;24:968-980.<br />
43. Carey RG, Teeters JL. CQI case study: reducing medication errors. Jt Comm J Qual<br />
Improv. 1995;21:232-237.<br />
44. Dinc L, Erdil F. The effectiveness <strong>of</strong> an educational intervention in changing nursing<br />
practice <strong>and</strong> preventing catheter-related infection for patients receiving total parenteral<br />
nutrition. Int J Nurs Stud. 2000;37:371-379.<br />
45. Cohran J, Larson E, Roach H, Blane C, Pierce P. Effect <strong>of</strong> intravascular surveillance <strong>and</strong><br />
education program on rates <strong>of</strong> nosocomial bloodstream infections. Heart Lung.<br />
1996;25:161-164.<br />
46. Flynn ER, Wolf ZR, McGoldrick TB, Jablonski RA, Dean LM, McKee EP. Effect <strong>of</strong> three<br />
teaching methods on a nursing staff's knowledge <strong>of</strong> medication error risk reduction<br />
strategies. J Nurs Staff Dev. 1996;12:19-26.<br />
47. Lundgren A, Wahren LK. Effect <strong>of</strong> education on evidence-based care <strong>and</strong> h<strong>and</strong>ling <strong>of</strong><br />
peripheral intravenous lines. J Clin Nurs. 1999;8:577-585.<br />
48. Gould D, Chamberlain A. The use <strong>of</strong> a ward-based educational teaching package to<br />
enhance nurses' compliance with infection control procedures. J Clin Nurs. 1997;6:55-67.<br />
49. Hendryx MS, Fieselmann JF, Bock MJ, Wakefield DS, Helms CM, Bentler SE. Outreach<br />
education to improve quality <strong>of</strong> rural ICU care. Results <strong>of</strong> a r<strong>and</strong>omized trial. Am J Respir<br />
Crit Care Med. 1998;158:418-23.<br />
440
50. Kite K. Changing mouth care practice in intensive care: implications <strong>of</strong> the clinical setting<br />
context. Intensive Crit Care Nurs. 1995;11:203-209.<br />
51. Knoblauch SC, Wilson CJ. Clinical outcomes <strong>of</strong> educating nurses about pediatric pain<br />
management. Outcomes Manag Nurs Pract. 1999;3:87-89.<br />
52. Spetz J, C<strong>of</strong>fman J. Maintaining an adequate supply <strong>of</strong> registered nurses in California: the<br />
case for increased public investment in nursing education [abstract]. Abstract Book /<br />
Association for Health Services Research. 1999:426-427.<br />
53. Barnason S, Merboth M, Pozehl B, Tietjen MJ. Utilizing an outcomes approach to improve<br />
pain management by nurses: a pilot study . Clin Nurse Spec. 1998;12:28-36.<br />
54. Fern<strong>and</strong>ez-Perez C, Tejada J, Carrasco M. Multivariate time series analysis in nosocomial<br />
infection surveillance: a case study. Int J Epidemiol. 1998;27:282-288.<br />
55. Goetz AM, Kedzuf S, Wagener M, Muder RR. Feedback to nursing staff as an intervention<br />
to reduce catheter-associated urinary tract infections. Am J Infect Control. 1999;27:402-<br />
404.<br />
56. Huda A, Wise LC. Evolution <strong>of</strong> compliance within a fall prevention program. J Nurs Care<br />
Qual. 1998;12:55-63.<br />
57. Moore SM, Wise L. Reducing nosocomial pressure ulcers. J Nurs Adm. 1997;27:28-34.<br />
58. Frykberg RG. The team approach in diabetic foot management. Adv Wound Care.<br />
1998;11:71-77.<br />
59. Gunningberg L, Lindholm C, Carlsson M, Sjoden PO. Implementation <strong>of</strong> risk assessment<br />
<strong>and</strong> classification <strong>of</strong> pressure ulcers as quality indicators for patients with hip fractures. J<br />
Clin Nurs. 1999;8:396-406.<br />
60. Rubenstein LZ, Robbins AS, Josephson KR, Schulman BL, Osterweil D. The value <strong>of</strong><br />
assessing falls in an elderly population. A r<strong>and</strong>omized clinical trial. Ann Intern Med.<br />
1990;113:308-316.<br />
61. Meier PA, Fredrickson M, Catney M, Nettleman MD. Impact <strong>of</strong> a dedicated intravenous<br />
therapy team on nosocomial bloodstream infection rates. Am J Infect Control.<br />
1998;26:388-392.<br />
62. Naylor M, Brooten D, Jones R, Lavizzo-Mourey R, Mezey M, Pauly M. Comprehensive<br />
discharge planning for the hospitalized elderly. A r<strong>and</strong>omized clinical trial. Ann Intern<br />
Med. 1994;120:999-1006.<br />
63. Pratt R, Burr G, Leelarthaepin B, Blizard P, Walsh S. The effects <strong>of</strong> All-RN <strong>and</strong> RN-EN<br />
staffing on the quality <strong>and</strong> cost <strong>of</strong> patient care. Aust J Adv Nurs. 1993;10:27-39.<br />
64. Ridsdale L, Kwan I, Cryer C. Newly diagnosed epilepsy: can nurse specialists help? A<br />
r<strong>and</strong>omized controlled trial. Epilepsy Care Evaluation Group. Epilepsia. 2000;41:1014-<br />
1019.<br />
65. York R, Brown LP, Samuels P, et al. A r<strong>and</strong>omized trial <strong>of</strong> early discharge <strong>and</strong> nurse<br />
specialist transitional follow-up care <strong>of</strong> high-risk childbearing women. Nursing Research.<br />
1997;46:254-261.<br />
66. Evans ML, Dick MJ, Shields DR, Shook DM, Smith MB. Postpartum sleep in the hospital.<br />
Relationship to taking-in <strong>and</strong> taking-hold. Clin Nurs Res. 1998;7:379-389.<br />
67. Mock V, Dow KH, Meares CJ, et al. Effects <strong>of</strong> exercise on fatigue, physical functioning,<br />
<strong>and</strong> emotional distress during radiation therapy for breast cancer. Oncol Nurs Forum.<br />
1997;24:991-1000.<br />
68. Sheely LC. Sleep disturbances in hospitalized patients with cancer. Oncol Nurs Forum.<br />
1996;23:109-111.<br />
441
69. ter Riet G, Kessels AG, Knipschild PG. R<strong>and</strong>omized clinical trial <strong>of</strong> ascorbic acid in the<br />
treatment <strong>of</strong> pressure ulcers. J Clin Epidemiol. 1995;48:1453-1460.<br />
70. Hardyck C, Petrinovich L. Reducing urinary tract infections in catheterized patients.<br />
Ostomy/Wound Management. 1998;44:36-8, 40, 42-43.<br />
71. Stotts NA, Barbour S, Griggs K, et al. Sterile versus clean technique in postoperative<br />
wound care <strong>of</strong> patients with open surgical wounds: a pilot study. J Wound Ostomy<br />
Continence Nurs. 1997;24:10-18.<br />
72. Wipke-Tevis DD, Stotts NA. Effect <strong>of</strong> dressings on saphenous vein harvest incision pain,<br />
distress <strong>and</strong> cosmetic result. Prog Cardiovasc Nurs. 1998;13:3-13.<br />
73. Gilcreast DM, Stotts NA, Froelicher ES, Baker LL, Moss KM. Effect <strong>of</strong> electrical<br />
stimulation on foot skin perfusion in persons with or at risk for diabetic foot ulcers. Wound<br />
Repair Regen. 1998;6:434-441.<br />
74. Spetz J, Seago JA, C<strong>of</strong>fman J, Rosen<strong>of</strong>f E, O'Neil E. Minimum Nurse Staffing Ratios in<br />
California Acute Care Hosptials. San Francisco, CA: University <strong>of</strong> California, San<br />
Francisco Center for the Health Pr<strong>of</strong>essions; 2000.<br />
75. Rogut L, Hudson A. Meeting patients' needs: quality care in a changing environment. Pap<br />
Ser United Hosp Fund N Y. 1995:1-33.<br />
76. Minion M, Ogden C, Brune D. <strong>Patient</strong> <strong>and</strong> staff needs drive changes on a postsurgical unit.<br />
Nurs Adm Q. 1994;18:18-28.<br />
77. Managed care <strong>and</strong> nursing: a view from the front lines. Massachusetts Nurse. 1995;65:4,<br />
10-11.<br />
78. Seago JA, Faucett J. Job strain among registered nurses <strong>and</strong> other hospital workers. J Nurs<br />
Adm. 1997;27:19-25.<br />
79. Seago JA. Evaluation <strong>of</strong> work redesign: <strong>Patient</strong> focused care. J Nurs Adm. 1999;29:31-38.<br />
80. Sochalski J, Aiken LH, Fagin C. Hospital Restructuring in the United States, Canada, <strong>and</strong><br />
Western Europe: An Outcomes Research Agenda. Med Care. 1997;35:OS13-25.<br />
81. Sochalski J, Boulis A, Shamian J, Buchan J, Muller-Mundt G. Lessons for restructuring the<br />
nursing workforce in North American <strong>and</strong> European hospitals [abstract]. Abstract Book /<br />
Association for Health Services Research. 1997;14:207-208.<br />
82. Weinberg DB. Why aare the nurses crying? Restructuring, power, an dcontrol in an<br />
American hospital. Boston, MA: Harvard; 2000.<br />
83. Blegen MA, Goode CJ, Reed L. Nurse staffing <strong>and</strong> patient outcomes. Nurs Res.<br />
1998;47:43-50.<br />
84. Blegen MA, Vaughn T. A multisite study <strong>of</strong> nurse staffing <strong>and</strong> patient occurrences.<br />
Nursing Economics. 1998;16:196-203.<br />
85. McCloskey JM. Nurse staffing <strong>and</strong> patient outcomes. Nursing Outlook. 1998;46:199-200.<br />
86. Reed L, Blegen MA, Goode CS. Adverse patient occurrences as a measure <strong>of</strong> nursing care<br />
quality. J Nurs Adm. 1998;28:62-69.<br />
87. Stevenson B, Mills EM, Welin L, Beal KG. Falls risk factors in an acute-care setting: a<br />
retrospective study. Can J Nurs Res. 1998;30:97-111.<br />
88. Sutton JC, St<strong>and</strong>en PJ, Wallace WA. <strong>Patient</strong> accidents in hospital: incidence,<br />
documentation <strong>and</strong> significance. Br J Clin Pract. 1994;48:63-66.<br />
89. Tutuarima JA, de Haan RJ, Limburg M. Number <strong>of</strong> nursing staff <strong>and</strong> falls: a case-control<br />
study on falls by stroke patients in acute-care settings. J Adv Nurs. 1993;18:1101-1105.<br />
90. Wolf ZR, McGoldrick TB, Flynn ER, Warwick F. Factors associated with a perceived<br />
harmful outcome from medication errors: a pilot study. J Contin Educ Nurs. 1996;27:65-<br />
74.<br />
442
91. Medicare. Available at: www.state.ia.us/ins/shiip/resources.html. Accessed June 23, 2001.<br />
92. Rhode Isl<strong>and</strong> Department <strong>of</strong> Health. Available at: http://www.health.state.ri.us//. Accessed<br />
June, 23 2001.<br />
93. California Consumer Health Scope. Available at: www.healthscope.org/hospital/where.asp.<br />
Accessed June 30, 2001.<br />
94. Denby C, Hogan N. Retaining Nursing Talent. In: Company TAB, ed. The Journey<br />
Begins. Atlanta, GA; 1999.<br />
95. Ballard K. Legislating <strong>Patient</strong> Advocacy: Nurse Staffing Ratios. In: Curtin L, ed. A<br />
Staffing Crisis: Nurse/<strong>Patient</strong> Ratios. Washington, DC; 2000.<br />
96. Ober S. Legislating <strong>Patient</strong> Advocacy: Nurse Staffing Ratios. In: Curtin L, ed. Staffing<br />
Crisis: Nurse/<strong>Patient</strong> Ratios. Washington, DC; 2000.<br />
97. 1997 NJSNA annual survey. New Jersey Nurse. 1997;27:7.<br />
98. Staffing is the top prioirty for MNA bargaining units. Massachusetts Nurse. 1999;69:5.<br />
99. MNA members take to the streets for safe patient care. Massachusetts Nurse. 1999;69:12.<br />
100. Staffing is the top priority for MNA bargaining units. Massachusetts Nurse. 1999;69:5.<br />
101. New study confirms critical role <strong>of</strong> low nurse-to-patient ratio. Massachusetts Nurse.<br />
1999;69:5.<br />
102. Maryl<strong>and</strong> RNs win big in a pact that sets patient-load lilmits. Am J Nurs. 1990;127: 134.<br />
103. Cape Cod Hospital RNs ratify l<strong>and</strong>mark agreement. Massachusetts Nurse. 1997;67: 1-2.<br />
104. Ratio ruckus. Modern Healthcare. 1998;28:48.<br />
105. Alex<strong>and</strong>er JW, Bauerschmidt AD. Implications for nursing administration <strong>of</strong> the<br />
relationship <strong>of</strong> technology <strong>and</strong> structure to quality <strong>of</strong> care. Nurs Adm Q. 1987;11:1-10.<br />
106. Currell R, Wainwright P, Urquhart C. Nursing record systems: effects on nursing practice<br />
<strong>and</strong> health care outcomes (Cochrane Review). In: The Cochrane Library, Issue 4, 2000.<br />
Oxford.<br />
107. Curtin LL. Nursing productivity: from data to definition. Nurs Manag . 1995;26:25, 28-29,<br />
32-36.<br />
108. Kovner C, Rosenfeld P. Practice <strong>and</strong> employment trends among nurse practitioners in New<br />
York State. J N Y State Nurses Assoc. 1997;28:4-8.<br />
109. Park HA, Park JH. Development <strong>of</strong> a computerized patient classification <strong>and</strong> staffing<br />
system. Stud Health Technol Inform. 1997;46:508-511.<br />
110. Spetz J. The measurement <strong>of</strong> technology in studies <strong>of</strong> the hospital industry [abstract]. Ahsr<br />
<strong>and</strong> Fhsr Annual Meeting Abstract Book. 1996;13:100.<br />
111. Mitchell PH, Shortell SM. Adverse outcomes <strong>and</strong> variations in organization <strong>of</strong> care<br />
delivery. Med Care. 1997;35:NS19-32.<br />
112. Brooten D, Roncoli M, Finkler S, Arnold L, Cohen A, Mennuti M. A r<strong>and</strong>omized trial <strong>of</strong><br />
early hospital discharge <strong>and</strong> home follow-up <strong>of</strong> women having cesarean birth. Obstet<br />
Gynecol. 1994;84:832-838.<br />
113. Alex<strong>and</strong>er JS, Younger RE, Cohen RM, Crawford LV. Effectiveness <strong>of</strong> a nurse-managed<br />
program for children with chronic asthma. J Pediatr Nurs. 1988;3:312-317.<br />
114. Jorgensen HS, Kammersgaard LP, Houth J, et al. Who benefits from treatment <strong>and</strong><br />
rehabilitation in a stroke Unit? A community-based study. Stroke. 2000;31:434-439.<br />
115. Aiken LH, Sloane DM, Lake ET, Sochalski J, Weber AL. Organization <strong>and</strong> outcomes <strong>of</strong><br />
inpatient AIDS care. Med Care. 1999;37:760-772.<br />
116. Hartz AJ, Krakauer H, Kuhn EM, et al. Hospital characteristics <strong>and</strong> mortality rates. N Engl<br />
J Med; 1989:1720-1725.<br />
443
117. Hunt J, Hagen S. Occasional paper. Nurse to patient ratios <strong>and</strong> patient outcomes. Nursing<br />
Times. 1998;94:63-66.<br />
118. Robertson RH, Hassan M. Staffing intensity, skill mix <strong>and</strong> mortality outcomes: the case <strong>of</strong><br />
chronic obstructive lung disease. Health Services Management Research. 1999;12:258-<br />
268.<br />
119. Hagen S, Hunt J. Investigating the relationship between nurse: patient ratios <strong>and</strong> patient<br />
outcomes using nationally collected data [abstract]. Abstract Book / Association for Health<br />
Services Research. 1998;15:103.<br />
120. Hunt J, Hagen S. Nurse to patient ratios <strong>and</strong> patient outcomes. Nurs Times. 1998;94:63-66.<br />
121. Needleman J, Buerhaus PI, Mattke S, Stewart M, Zelevinsky K. Nurse Staffing <strong>and</strong> <strong>Patient</strong><br />
Outcomes in Hospitals. Boston, MA: US Department <strong>of</strong> Health <strong>and</strong> Human Services,<br />
Health Resources <strong>and</strong> Service Administration; 2001.<br />
122. American Nurses Association Network Inc. Implementing nursing's report card: A study<br />
<strong>of</strong> RN staffing, length <strong>of</strong> stay, <strong>and</strong> patient outcomes. Washington, DC: American Nurses<br />
Publishing; 1997:32.<br />
123. American Nurses Association Network Inc. Nurse Staffing <strong>and</strong> <strong>Patient</strong> Outcomes in the<br />
Inpatient Hospital Setting. Washington, DC: American Nurses Publishing; 2000:53.<br />
124. Kovner C, Gergen PJ. Nurse staffing levels <strong>and</strong> adverse events following surgery in U.S.<br />
hospitals. Image J Nurs Sch. 1998;30:315-321.<br />
125. Lichtig LK, Knauf RA, Milholl<strong>and</strong> DK. Some impacts <strong>of</strong> nursing on acute care hospital<br />
outcomes. J Nurs Adm. 1999;29:25-33.<br />
126. Fridkin SK, Pear SM, Williamson TH, Galgiani JN, Jarvis WR. The role <strong>of</strong> understaffing<br />
in central venous catheter-associated bloodstream infections. Infect Control Hosp<br />
Epidemiol. 1996;17:150-158.<br />
127. Flood SD, Diers D. Nurse staffing, patient outcome <strong>and</strong> cost. Nurs Manag . 1988;19:34-35,<br />
38-39, 42-43.<br />
128. Kuhn EM, Hartz AJ, Gottlieb MS, Rimm AA. The relationship <strong>of</strong> hospital characteristics<br />
<strong>and</strong> the results <strong>of</strong> peer review in six large states. Med Care. 1991;29:1028-1038.<br />
129. Wan TT, Shukla RK. Contextual <strong>and</strong> organizational correlates <strong>of</strong> the quality <strong>of</strong> hospital<br />
nursing care. QRB Qual Rev Bull. 1987;13:61-64.<br />
130. Grillo-Peck AM, Risner PB. The Effect <strong>of</strong> a Partnership Model on Quality <strong>and</strong> Length <strong>of</strong><br />
Stay. Nursing Economics. 1995;13:367-374.<br />
131. Pierce SF. Nurse-sensitive health care outcomes in acute care settings: an integrative<br />
analysis <strong>of</strong> the literature. Journal <strong>of</strong> Nursing Care Quality; 1997:60-72.<br />
132. Czaplinski C, Diers D. The effect <strong>of</strong> staff nursing on length <strong>of</strong> stay <strong>and</strong> mortality. Med<br />
Care. 1998;36:1626-1638.<br />
133. Aiken LH, Sloane DM. Effects <strong>of</strong> specialization <strong>and</strong> client differentiation on the status <strong>of</strong><br />
nurses: the case <strong>of</strong> AIDS. J Health Soc Behav. 1997;38:203-222.<br />
134. Aubert RE, Herman WH, Waters J, et al. Nurse case management to improve glycemic<br />
control in diabetic patients in a health maintenance organization. A r<strong>and</strong>omized, controlled<br />
trial. Ann Intern Med. 1998;129:605-612.<br />
135. Borkowski V. Implementation <strong>of</strong> a managed care model in an acute care setting. J Healthc<br />
Qual. 1994;16:25-27, 30.<br />
136. Clark JM, Maben J. Health promotion in primary health care nursing: the development <strong>of</strong><br />
quality indicators. Health Educ J. 1999;58:99-119.<br />
137. Dring R, Hiott B, Elliott K. Case management: a case study. J Neurosci Nurs.<br />
1994;26:166-169.<br />
444
138. Jones J, Rowan K. Is there a relationship between the volume <strong>of</strong> work carried out in<br />
intensive care <strong>and</strong> its outcome? Int J Technol Assess Health Care. 1995;11:762-769.<br />
139. Strzalka A, Havens DS. Nursing care quality: comparison <strong>of</strong> unit-hired, hospital float pool,<br />
<strong>and</strong> agency nurses. J Nurs Care Qual. 1996;10:59-65.<br />
140. Westcott E, Dunn V. An exploration <strong>of</strong> the value <strong>of</strong> specialist neurosurgical nurses...<br />
including commentary by Northcott N. NT Research. 1998;3:421-431.<br />
141. Gilbert M, Counsell C. Planned change to implement a restraint reduction program J Nurs<br />
Care Qual. 1999;13:57-64. Published erratum appears in J Nurs Care Qual 1999;13.<br />
142. von Koch L, Widen Holmqvist L, Kostulas V, Almazan J, de Pedro-Cuesta J. A<br />
r<strong>and</strong>omized controlled trial <strong>of</strong> rehabilitation at home after stroke in Southwest Stockholm:<br />
outcome at six months. Sc<strong>and</strong> J Rehabil Med. 2000;32:80-86.<br />
143. Wagner EH, LaCroix AZ, Grothaus L, et al. Preventing disability <strong>and</strong> falls in older adults:<br />
a population-based r<strong>and</strong>omized trial. Am J <strong>Public</strong> Health. 1994;84:1800-1806.<br />
144. Day G, Hindmarsh J, Hojna C, Roy G, Ventimiglia N. Improving medication<br />
administration through an enhanced occurrence reporting system. J Nurs Care Qual.<br />
1994;9:51-56.<br />
145. Mendelson MH, Short LJ, Schechter CB, et al. Study <strong>of</strong> a needleless intermittent<br />
intravenous-access system for peripheral infusions: analysis <strong>of</strong> staff, patient, <strong>and</strong><br />
institutional outcomes. Infect Control Hosp Epidemiol. 1998;19:401-406.<br />
146. Rush J, Fiorino-Chiovitti R, Kaufman K, Mitchell A. A r<strong>and</strong>omized controlled trial <strong>of</strong> a<br />
nursery ritual: wearing cover gowns to care for healthy newborns. Birth. 1990;17:25-30.<br />
147. Gilbert B. Facts, not myths. Study shows for-pr<strong>of</strong>it hospital conversions help communities<br />
by injecting capital. Mod Healthc; 1997:54.<br />
148. Moore MM, Nguyen D, Nolan SP, et al. Interventions to reduce decibel levels on patient<br />
care units. Am Surg. 1998;64:894-899.<br />
149. Spetz J. Hospital use <strong>of</strong> nursing personnel: holding steady through the 1990s. J Nurs Adm.<br />
2000;30:344-346.<br />
150. C<strong>of</strong>fman J, Spetz J. Maintaining an adequate supply <strong>of</strong> RNs in California. Image J Nurs<br />
Sch. 1999;31:389-393.<br />
151. C<strong>of</strong>fman J, Spetz J, Seago JA, Rosen<strong>of</strong>f E, O;Neil E. Nursing in California: A Workforce<br />
in Crisis. . San Francisco, CA: UCSF Center for the Health Pr<strong>of</strong>essions; 2001:91.<br />
152. Aiken LH, Smith HL, Lake ET. Lower medicare mortality among a set <strong>of</strong> hospitals known<br />
for good nursing care. Med Care. 1994;32:771-787.<br />
153. Aiken LH, Clarke SP, Sloane DM. Hospital Restructuring: Does it Adversely Affect Care<br />
<strong>and</strong> Outcomes? J Health Hum Serv Adm. 2000; In press.<br />
154. Archibald LK, Manning ML, Bell LM, Banerjee S, Jarvis WR. <strong>Patient</strong> density, nurse-topatient<br />
ratio <strong>and</strong> nosocomial infection risk in a pediatric cardiac intensive care unit. Pediatr<br />
Infect Dis J. 1997;16:1045-1048.<br />
155. Harbarth S, Sudre P, Dharan S, Cadenas M, Pittet D. Outbreak <strong>of</strong> Enterobacter cloacae<br />
related to understaffing, overcrowding, <strong>and</strong> poor hygiene practices. Infect Control Hosp<br />
Epidemiol. 1999;20:598-603.<br />
156. Robert J, Fridkin SK, Blumberg HM, et al. The influence <strong>of</strong> the composition <strong>of</strong> the nursing<br />
staff on primary bloodstream infection rates in a surgical intensive care unit. Infect Control<br />
Hosp Epidemiol. 2000;21:12-17.<br />
157. Taunton RL, Kleinbeck SV, Stafford R, Woods CQ, Bott MJ. <strong>Patient</strong> outcomes. Are they<br />
linked to registered nurse absenteeism, separation, or work load? J Nurs Adm. 1994;24:48-<br />
55.<br />
445
158. Amaravadi RK, Dimick JB, Pronovost PJ, Lipsett PA. ICU nurse-to-patient ratio is<br />
associated with complications <strong>and</strong> resource use after esophagectomy. Intensive Care Med.<br />
2000;26:1857-1862.<br />
159. Silber JH, Rosenbaum PR, Ross RN. Comparing the contributions <strong>of</strong> groups <strong>of</strong> predictors:<br />
which outcomes vary with hospital rather than patient characteristics? Journal<strong>of</strong> the<br />
American Statistical Association. 1995;90:7-18.<br />
160. Zimmerman JE, Shortell SM, Knaus WA, et al. Value <strong>and</strong> cost <strong>of</strong> teaching hospitals: a<br />
prospective, multicenter, inception cohort study . Crit Care Med. 1993;21:1432-1442.<br />
161. Wan TT. Hospital variations in adverse patient outcomes. Qual Assur Util Rev. 1992;7:50-<br />
53.<br />
162. Shukla R. Structure vs. people in primary nursing: An inquiry. Nursing Research.<br />
1981;30:236-241.<br />
446
Chapter 40. Promoting a Culture <strong>of</strong> <strong>Safety</strong><br />
Laura T. Pizzi, PharmD<br />
Neil I. Goldfarb<br />
David B. Nash, MD, MBA<br />
Thomas Jefferson University School <strong>of</strong> Medicine<br />
<strong>and</strong> Office <strong>of</strong> Health Policy & Clinical Outcomes<br />
Background<br />
In a number <strong>of</strong> high hazard organizations, where the risk <strong>of</strong> error involves dire<br />
consequences, leaders manage for safe, reliable performance. As a result, the term High<br />
Reliability Organization has been coined to describe organizations with exemplary track records<br />
<strong>of</strong> safety: aviation, chemical manufacturing, shipping, nuclear power production, <strong>and</strong> the<br />
military. 1-10 This concept is rooted in the analyses <strong>of</strong> errors that reveal organizational failures,<br />
along with technical failures (related to system performance) <strong>and</strong> human limitations (related to<br />
human behavior). 11<br />
Theories about antecedents to accidents abound but major schools <strong>of</strong> thought include<br />
Reason’s belief that a number <strong>of</strong> latent factors embedded in organizational systems can align <strong>and</strong><br />
result in accidents, 12-14 <strong>and</strong> Rasmussen’s approach to categorizing the different sources <strong>of</strong> error<br />
that interact with latent factors to produce accidents. 15-17 Another school <strong>of</strong> thought, developed<br />
by Charles Perrow <strong>and</strong> first publicized shortly after the Three Mile Isl<strong>and</strong> nuclear accident,<br />
Normal Accident Theory, 18,19 emphasizes the ever-present possibility <strong>of</strong> accidents in<br />
organizations that exhibit complexity <strong>and</strong> “tight coupling” <strong>of</strong> processes <strong>and</strong> the inevitability <strong>of</strong><br />
accidents. Normal Accident Theory st<strong>and</strong>s somewhat in opposition to High Reliability Theory,<br />
which holds that accidents can be prevented through organizational design <strong>and</strong> management.<br />
Scott Sagan’s analysis <strong>of</strong> the nuclear weapons industry, addressing the question <strong>of</strong> why there has<br />
never been an ‘accidental’ nuclear war, represents a fascinating investigation <strong>of</strong> a test case for<br />
these two schools <strong>of</strong> thought. 20 Despite the obvious apparent confirmation <strong>of</strong> the High Reliability<br />
Theory perspective (ie, such an accident has thankfully never occurred), Sagan uncovers a<br />
surprising amount <strong>of</strong> evidence that also seems to confirm the Normal Accident perspective.*<br />
Regardless <strong>of</strong> the underlying theory, health care is vulnerable to error. The application <strong>of</strong><br />
safety promotion theories utilized to positive effect in other high hazard organizations are being<br />
considered for health care, where “accidents” tend to occur one person at a time instead <strong>of</strong> in<br />
sweeping disasters. 25<br />
Attention to organizational issues <strong>of</strong> structure, strategy <strong>and</strong> culture may be a promising<br />
direction for medicine. Although organizational elements are intertwined <strong>and</strong> must be aligned for<br />
optimum performance 26 this chapter focuses on the culture component, especially “safety<br />
cultures.” Following a description <strong>of</strong> the prevailing models <strong>of</strong> culture <strong>and</strong> safety, we review<br />
approaches that both medical <strong>and</strong> non-medical industries have used to promote a culture <strong>of</strong><br />
safety. On the medical side, the discussion is limited to the Veterans Health Administration’s<br />
comprehensive safety initiative. 27 On the non-medical side, specific methods other high<br />
* Invited commentaries on Diane Vaughn’s in-depth analysis <strong>of</strong> the Challenger crash 21 also<br />
provide an interesting comparison between the Normal Accidents 22 <strong>and</strong> High Reliability Theory<br />
23,24 perspectives <strong>and</strong> indicates that they are more complementary than contradictory.<br />
447
eliability industries have applied to promote a safety culture, 4 including a behavior-based<br />
industry approach, are reported. 28<br />
Organizational Culture<br />
Helmreich defines culture as “a complex framework <strong>of</strong> national, organizational, <strong>and</strong><br />
pr<strong>of</strong>essional attitudes <strong>and</strong> values within which groups <strong>and</strong> individuals function.” 29 Corporate<br />
culture is <strong>of</strong>ten referred to as the glue that holds an organization together, <strong>and</strong> is therefore<br />
assumed to be a contributor to organizational performance by socializing workers in a way that<br />
increases commitment to the goals <strong>of</strong> the entity. 4,30,31 As such, it embodies the philosophy <strong>of</strong><br />
senior leaders, which is translated into, <strong>and</strong> affects the behaviors <strong>of</strong> employees. 32 Although some<br />
schools <strong>of</strong> thought focus on the role <strong>of</strong> leaders <strong>of</strong> an organization (board members <strong>and</strong><br />
executives), others note that middle management likely plays a substantial role as well,<br />
conveying the culture to front-line workers in any organization, as evidenced by studies <strong>of</strong> the<br />
effective use <strong>of</strong> total quality management. 33 The power <strong>of</strong> culture <strong>of</strong>ten goes unrecognized, since<br />
employees may assume that the dominant paradigm is simply “the way we do things here.” 29<br />
<strong>Safety</strong> Culture<br />
While an exact definition <strong>of</strong> a safety culture does not exist, a recurring theme in the<br />
literature is that organizations with effective safety cultures share a constant commitment to<br />
safety as a top-level priority, which permeates the entire organization. More concretely, noted<br />
components include: 1) acknowledgment <strong>of</strong> the high risk, error-prone nature <strong>of</strong> an organization’s<br />
activities, 2) blame-free environment where individuals are able to report errors or close calls<br />
without punishment, 3) expectation <strong>of</strong> collaboration across ranks to seek solutions to<br />
vulnerabilities, <strong>and</strong> 4) willingness on the part <strong>of</strong> the organization to direct resources to address<br />
safety concerns. 3,4,29,34-36 Based on extensive field work in multiple organizations, Roberts et al<br />
have observed several common, cultural values in reliability enhancing organizations:<br />
“interpersonal responsibility; person centeredness; [co-workers] helpful <strong>and</strong> supportive <strong>of</strong> one<br />
another; friendly, open sensitive personal relations; creativity; achieving goals, strong feelings <strong>of</strong><br />
credibility; strong feelings <strong>of</strong> interpersonal trust; <strong>and</strong> resiliency.” 4<br />
Culture Surveys<br />
The aspect <strong>of</strong> organizational safety culture that may be visible or measurable is<br />
sometimes referred to as the safety “climate,” which includes management systems, safety<br />
systems, <strong>and</strong> individual attitudes <strong>and</strong> perceptions. 32 Health care organizations are now adapting<br />
safety culture <strong>and</strong> climate surveys from other industries to benchmark <strong>and</strong> identify potential<br />
deficiencies in their unique safety culture. Kaiser Permanente, the oldest <strong>and</strong> largest not-forpr<strong>of</strong>it<br />
health maintenance organization in the United States, has administered an executive<br />
attitudes <strong>and</strong> beliefs survey to identify perceptions <strong>of</strong> patient safety for the purposes <strong>of</strong> planning<br />
<strong>and</strong> measurement (written communication, February 2001, Suzanne Graham). The VA Palo Alto<br />
<strong>Patient</strong> <strong>Safety</strong> Center <strong>of</strong> Inquiry <strong>and</strong> Stanford University’s Center for Health Policy/Center for<br />
Primary Care <strong>and</strong> Outcomes Research are conducting a patient safety culture survey that builds<br />
on past work by Gaba <strong>and</strong> collaborators. The survey includes items on production pressures <strong>and</strong><br />
safety consequences, <strong>and</strong> draws from several other sources (personal communication, June,<br />
2001, Sara Singer). Spath provides a checklist <strong>of</strong> elements that health care managers can use to<br />
identify which cultural elements should be addressed in order to improve safety 37 (Table 40.1).<br />
Previous work in assessing organizational culture effects on total quality management, 38 <strong>and</strong><br />
448
organizational culture in high reliability organizations 39 may also be pertinent to efforts to<br />
measure culture <strong>and</strong> its consequences for patient safety.<br />
Industries Outside Medicine<br />
Promoting a culture <strong>of</strong> safety has historically been a priority for the chemical, electrical,<br />
food processing, petroleum, plastic, <strong>and</strong> transportation industries. Since the 1930s, safety<br />
managers within various industries have recognized that most occupational injuries have a strong<br />
behavioral component, typically rooted in the safety culture. 28 In these settings, behavior<br />
analysis has been used as an approach to solving safety problems. Behavioral analyses typically<br />
involve assessing upstream <strong>and</strong> downstream behaviors associated with the problem, with further<br />
analysis as to which behaviors may be modifiable. Once relevant behaviors are identified, a<br />
behavior change intervention is implemented, <strong>and</strong> behavioral changes are measured.<br />
Interventions are customized, <strong>and</strong> draw upon techniques <strong>of</strong> behavior science, organizational<br />
development, safety science, <strong>and</strong> quality. Researchers have shown associations between<br />
behavior-based safety programs <strong>and</strong> reduced rates <strong>of</strong> accidents.<br />
In an extensive field study <strong>of</strong> three organizations (nuclear aircraft carriers, a nuclear<br />
power plant, <strong>and</strong> the Federal agency responsible for air traffic control) whose operations have<br />
the potential for widespread harm, Roberts et al proposed several management processes that<br />
“cradle” a culture <strong>of</strong> perfection. 4 One process requires distributing decision making, while<br />
having mechanisms that allow decisions to migrate up <strong>and</strong> down the chain <strong>of</strong> comm<strong>and</strong> as<br />
circumstances develop. The mechanism for localizing decision making is <strong>of</strong>ten extensive<br />
training, while the approach to moving decisions to higher levels is based on management by<br />
exception when acceptable operation is in question. Finally, these researchers suggest that both<br />
top-level managers <strong>and</strong> local operators develop a deep underst<strong>and</strong>ing <strong>of</strong> their organizations, <strong>and</strong><br />
use this “big picture” perspective to provide intuitive judgments when situations arise.<br />
Practice Description<br />
Veterans Health Administration Approach<br />
The Veterans Health Administration (VHA) has implemented a multifaceted safety<br />
initiative, which was designed to build a culture <strong>of</strong> safety <strong>and</strong> address system failures. 27 The<br />
approach consists <strong>of</strong> 4 major elements: 1) partnering with other safety-related organizations <strong>and</strong><br />
affiliates to demonstrate a public commitment by leadership, 2) establishing centers to direct<br />
safety efforts, 3) improving reporting systems, <strong>and</strong> 4) providing incentives to health care team<br />
members <strong>and</strong> division leaders. These tactics are detailed below. In addition, several specific<br />
initiatives were implemented to address problems, such as bar coding <strong>of</strong> medications<br />
(Subchapter 43.1) <strong>and</strong> use <strong>of</strong> computerized medical records.<br />
To demonstrate a public commitment to the importance <strong>of</strong> patient safety, the VHA<br />
leadership founded the National <strong>Patient</strong> <strong>Safety</strong> Partnership, along with several major healthrelated<br />
organizations (the American Association <strong>of</strong> <strong>Medical</strong> Colleges, the American Hospital<br />
Association, the American <strong>Medical</strong> Association, the American Nurses Association, <strong>and</strong> the<br />
Institute for Healthcare Improvement). In addition, key senior management <strong>of</strong>ficials sounded the<br />
safety message in congressional testimony.<br />
The second part <strong>of</strong> the VHA's approach involved establishing centers dedicated to the<br />
promotion <strong>of</strong> patient safety. The first <strong>of</strong> these was the National Center for <strong>Patient</strong> <strong>Safety</strong>, which<br />
directs patient safety efforts for the VHA at a national level. The Director <strong>of</strong> the Center oversees<br />
patient safety efforts for the entire VHA health system <strong>and</strong> is a recognized authority.<br />
449
Subsequently, four <strong>Patient</strong> <strong>Safety</strong> Centers for Inquiry were funded, which are primarily<br />
responsible for safety-related research <strong>and</strong> development. Specifically, the centers are responsible<br />
for identifying problems in the patient care process, implementing corrective measures, <strong>and</strong><br />
studying effects. Currently, one <strong>of</strong> these centers is studying safety cultures in health care<br />
organizations. Finally, the VHA's Virtual Learning Center contributes to the safety initiative by<br />
allowing VHA facilities to share lessons learned. Additional information, such as training,<br />
educational programs, alerts, <strong>and</strong> advisories are planned.<br />
The third major component <strong>of</strong> the VHA's initiative involves incentives aimed at<br />
improving safety. There are two types <strong>of</strong> incentives <strong>of</strong>fered: 1) the "carrot," which is a monetary<br />
award <strong>of</strong> up to $5000 for individuals <strong>and</strong> teams that develop approaches to improve safety<br />
issues; <strong>and</strong> 2) the "stick," which is a performance expectation imposed on leaders to improve<br />
patient safety. Leaders <strong>of</strong> the VHA's 22 regional networks must demonstrate involvement in<br />
safety-promoting activities, or be subject to consequences, including possible termination <strong>of</strong><br />
employment. The primary objective <strong>of</strong> this incentive is to align regional <strong>and</strong> national leaders'<br />
goals.<br />
Last, the VHA has implemented a two-pronged system for capturing adverse events. The<br />
first <strong>of</strong> these systems, the <strong>Patient</strong> <strong>Safety</strong> Event Registry, m<strong>and</strong>ates reporting <strong>of</strong> adverse events<br />
<strong>and</strong> “close calls” occurring within the system. Before implementing the <strong>Patient</strong> <strong>Safety</strong> Event<br />
Registry, regional review <strong>of</strong> event cases was sporadic. After implementation, event data is<br />
systematically shared both regionally <strong>and</strong> nationally. The second <strong>of</strong> the systems, the voluntary<br />
reporter identity system, was developed in conjunction with the National Aeronautics <strong>and</strong> Space<br />
Administration, <strong>and</strong> allows for anonymous event reporting. It is intended that the use <strong>of</strong> both<br />
reporting systems will together provide a more comprehensive picture <strong>of</strong> safety management<br />
than would be possible with one system alone.<br />
Behavior-Based <strong>Safety</strong> Programs<br />
Outside <strong>of</strong> medicine, the objective <strong>of</strong> behavior-based safety interventions is to reduce<br />
incidents by managing at-risk behaviors <strong>of</strong> the organization <strong>and</strong> work teams. An approach<br />
described by Krause <strong>and</strong> colleagues consisted <strong>of</strong> safety assessments, steering committee<br />
formation, development <strong>of</strong> checklists <strong>of</strong> well-specified critical behaviors related to safety,<br />
observer training regarding the critical behaviors, observation <strong>and</strong> feedback. 28 These steps,<br />
somewhat analogous to aspects <strong>of</strong> crew resource management training approaches (see Chapter<br />
44), most likely reflect an active safety culture. The Krause study assessed the effectiveness <strong>of</strong><br />
behavioral safety initiatives in reducing accidents in 229 facilities in various industries, including<br />
chemical, electrical, food, plastic, petroleum, transportation, service, <strong>and</strong> paper manufacturers. 28<br />
The study used an interrupted time series design with the participating industrial sites. Event<br />
rates after implementation <strong>of</strong> the behavioral program were compared with the Occupational<br />
<strong>Safety</strong> <strong>and</strong> Health Administration (OSHA) recordable illness/injury rates. Of the 229<br />
participating sites, 73 provided necessary data (others were excluded either because they failed<br />
to provide OSHA illness/injury rates or results <strong>of</strong> the behavioral initiative). Compared with<br />
baseline, the behavioral initiative resulted in an average 26% improvement in targeted safety<br />
behaviors during the first year, which rose to 69% by the fifth year.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
There is no known information about the prevalence <strong>of</strong> medical error emanating from<br />
cultural/organizational problems in health care. Culture is known to contribute to the occurrence<br />
<strong>of</strong> errors <strong>and</strong> accidents. Its contribution relative to other causal factors is unknown, but likely to<br />
450
vary, depending on the type <strong>of</strong> accident <strong>and</strong> work environment. 3,7,29 The aviation industry<br />
attributes its successful safety record in part to analysis <strong>of</strong> near miss <strong>and</strong> accident reports (see<br />
Chapter 4). 40-43 These types <strong>of</strong> analyses are only possible if the culture supports reporting <strong>of</strong><br />
errors. Culture changes may, in fact, have their greatest impact on “underground” (unreported)<br />
errors, which are extremely difficult to quantify.<br />
Opportunities for Impact<br />
Although no data from ongoing surveys has yet emerged to permit us to accurately<br />
quantify safety culture penetration, we nonetheless speculate based on anecdotal evidence that<br />
health care organizations have plenty <strong>of</strong> room for improvement. A number <strong>of</strong> observers have<br />
noted large-scale obstacles to promotion <strong>of</strong> safety culture within health care: a pervasive culture<br />
<strong>of</strong> blame that impedes acknowledgment <strong>of</strong> error, <strong>and</strong> pr<strong>of</strong>essional “silos” that <strong>of</strong>fer unique<br />
challenges to changing any universal aspect <strong>of</strong> health care, including culture. 44-46<br />
Even before the Institute <strong>of</strong> Medicine’s pivotal To Err is Human report was delivered to<br />
the public, promoting a safety culture within health care had received widespread attention. The<br />
Institute for Healthcare Improvement’s Web site features a report “Reducing <strong>Medical</strong> Errors <strong>and</strong><br />
Improving <strong>Patient</strong> <strong>Safety</strong>: Success Stories from the Front Lines <strong>of</strong> Medicine.” 47 It includes<br />
articles about the transformation <strong>of</strong> culture at the prestigious Dana-Farber Cancer Institute after a<br />
highly publicized chemotherapy overdose in 1994, which resulted in the death <strong>of</strong> a patient.<br />
Another article in the same series highlighted the major steps, including cultural change, as noted<br />
above, taken by leaders <strong>of</strong> the nation’s largest health care provider—the Veterans Affairs<br />
Healthcare System—after fatal medical errors were reported by the media. 47<br />
Comment<br />
Measuring the impact <strong>of</strong> culture on safety-related outcomes is challenging. Culture is a<br />
complex <strong>and</strong> abstract construct that must be inferred from behaviors, <strong>and</strong> analysis <strong>of</strong>ten relies on<br />
self-reported data. 29 Research continues to develop a working model <strong>of</strong> safety culture that<br />
permits measurement <strong>of</strong> several connected concepts: individuals’ perceptions <strong>and</strong> attitudes about<br />
safety, individuals’ observable safety behaviors, <strong>and</strong> an organization’s safety management<br />
system as evidenced by its policies <strong>and</strong> management styles. 35 The relative impact <strong>of</strong> each <strong>of</strong><br />
these measures on outcomes is another layer <strong>of</strong> ongoing research.<br />
Although some data support the effectiveness <strong>of</strong> the entire VHA initiative in improving<br />
safety, there are no direct data supporting the effect <strong>of</strong> promoting a culture <strong>of</strong> safety. The use <strong>of</strong><br />
incentives to reward safety-promoting behavior <strong>and</strong> publicly demonstrating a commitment to<br />
safety are approaches that could be applied in both large <strong>and</strong> small health care settings. The<br />
VHA’s reporting system will likely be watched, <strong>and</strong> potentially adapted by large providers who<br />
have inconsistent <strong>and</strong>/or insufficient reporting <strong>of</strong> safety problems at local, regional, <strong>and</strong> national<br />
levels.<br />
The evidence presented by Krause provides compelling support for the effectiveness <strong>of</strong><br />
behavior-based safety programs in a wide range <strong>of</strong> industrial settings. Although this exact<br />
approach has not been evaluated in health care environments, its emphasis on promoting safety<br />
culture seems applicable to patient care environments.<br />
As noted in To Err is Human, researchers who have studied organizations with a strong<br />
safety culture believe that it is “the most critical underlying feature <strong>of</strong> their accomplishments.” 48<br />
Although the nature <strong>of</strong> the evidence is based on field studies <strong>and</strong> other methods not typical <strong>of</strong><br />
medical evidence, it is considered compelling by a number <strong>of</strong> experts from organizational <strong>and</strong><br />
other social sciences. At this point, promoting a culture <strong>of</strong> safety remains surprisingly<br />
451
unexplored in health care settings, where the risks <strong>of</strong> error are high. Further research in this area<br />
is warranted, though the threshold for evidence may need a different yardstick than is typically<br />
applied in medicine (Chapter 2).<br />
Acknowledgments<br />
The authors are grateful to Mr. Marcus Moore <strong>of</strong> Behavioral Science Technology, Inc.<br />
for providing information related to industrial safety programs. We would also like to thank Dr.<br />
Kasey Thompson, Director <strong>of</strong> the American Society <strong>of</strong> Health-System Pharmacy’s Center for<br />
<strong>Patient</strong> <strong>Safety</strong>, <strong>and</strong> Mr. Mark Thomas, Director Of Pharmacy at Children’s Hospital in<br />
Minneapolis for their suggestions <strong>and</strong> insight on cultural safety issues.<br />
452
Table 40.1. Checklist <strong>of</strong> elements that contribute to a patient-safe environment<br />
‰ All people acknowledge that top management provides essential patient safety improvement leadership.<br />
‰ The organization has clearly defined patient safety polices.<br />
‰ All people can explain the organization’s patient safety policies.<br />
‰ All people are involved in developing patient safety goals, <strong>and</strong> everyone can explain desired<br />
results <strong>and</strong> measures.<br />
‰ All people are actively involved in identifying <strong>and</strong> resolving patient safety concerns.<br />
‰ All people can explain how their personal performance affects patient safety.<br />
‰ All people believe they have the necessary authority <strong>and</strong> resources to meet their<br />
responsibilities for patient safety<br />
‰ <strong>Patient</strong> safety performance for all people is measured against goal, clearly displayed, <strong>and</strong><br />
rewarded.<br />
‰ A comprehensive review <strong>of</strong> patient safety is conducted annually, <strong>and</strong> there is a process in<br />
place that drives continuous improvement.<br />
‰ Regular workplace hazard analyses are conducted to identify patient safety improvement<br />
opportunities. The results are used to make changes in patient care activities.<br />
‰ All people are empowered to correct patient safety hazards as they are identified.<br />
‰ A comprehensive system exists for gathering information on patient safety hazards. The<br />
system is positive, rewarding, <strong>and</strong> effective, <strong>and</strong> people use it.<br />
‰ All people are fully aware <strong>of</strong> patient incident trends, causes, <strong>and</strong> means <strong>of</strong> prevention.<br />
‰ All injury-producing patient incidents <strong>and</strong> significant “near misses” are investigated for root<br />
cause, with effective preventive actions taken.<br />
‰ All people who operate patient care equipment are trained to recognize maintenance needs<br />
<strong>and</strong> perform or request timely maintenance.<br />
‰ All people know immediately how to respond to an emergency because <strong>of</strong> effective planning,<br />
training, <strong>and</strong> drills.<br />
‰ Facilities are fully equipped for emergencies; all necessary systems <strong>and</strong> equipment are in<br />
place <strong>and</strong> regularly tested; <strong>and</strong> all people know how to use equipment <strong>and</strong> communicate<br />
during emergencies.<br />
‰ Ergonomics experts are provided when needed <strong>and</strong> are involved in patient safety assessment<br />
<strong>and</strong> training.<br />
453
‰ All supervisors/managers assist in patient safety workplace analyses, ensure physical<br />
protections, reinforce training, enforce discipline, <strong>and</strong> can explain how to provide safe<br />
patient care.<br />
454
References<br />
1. Roberts KH, Gargano G. Managing a high reliability organization: A case for<br />
interdependence. In: Glinow MAV, Mohrmon S, editors. Managing complexity in high<br />
technology industries: Systems <strong>and</strong> people. New York: Oxford University Press.; 1989:<br />
147-59.<br />
2. Gaba DM. Structural <strong>and</strong> organizational issues in patient safety: A comparison <strong>of</strong> health<br />
care to other high-hazard industries. California Management Review. 2000; 43: 83-102.<br />
3. Roberts KH. Some characteristics <strong>of</strong> high reliability organizations. Berkeley, CA ::<br />
Produced <strong>and</strong> distributed by Center for Research in Management, University <strong>of</strong> California,<br />
Berkeley Business School; 1988.<br />
4. Roberts KH. Cultural characteristics <strong>of</strong> reliability enhancing organizations. Journal <strong>of</strong><br />
Managerial Issues. 1993; 5: 165-81.<br />
5. Roberts KH. Managing high reliability organizations. California Management Review.<br />
1990; 32: 101-13.<br />
6. Roberts KH, Stout SK, Halpern JJ. Decision dynamics in two high reliability military<br />
organizations. Management Science. 1994; 40: 614-24.<br />
7. Weick KE. Organizational culture as a source <strong>of</strong> high reliability. California management<br />
review. 1987; 29: 112-27.<br />
8. Roberts KH, Libuser C. From Bhopal to banking: Organizational design can mitigate risk.<br />
Organizational Dynamics. 1993; 21: 15-26.<br />
9. LaPorte TR, Consolini P. Theoretical <strong>and</strong> operational challenges <strong>of</strong> "high-reliability<br />
organizations": Air-traffic control <strong>and</strong> aircraft carriers. International Journal <strong>of</strong> <strong>Public</strong><br />
Administration. 1998; 21: 847-52.<br />
10. LaPorte TR. The United States air traffic control system : increasing reliability in the midst<br />
<strong>of</strong> rapid growth. In: Mayntz R, Hughes TP, editors. The Development <strong>of</strong> Large technical<br />
Systems. Boulder, CO:: Westview Press; 1988.<br />
11. Van Vuuren W. Organizational failure: lessons from industry applied to the medical<br />
domain. <strong>Safety</strong> Science. 1999; 33: 14-29.<br />
12. Reason JT. Human Error. New York: Cambridge Univ Press; 1990.<br />
13. Reason JT. Managing the Risks <strong>of</strong> Organizational Accidents: Ashgate Publishing<br />
Company; 1997.<br />
14. Reason J. Human error: models <strong>and</strong> management. BMJ. 2000; 320: 768-70.<br />
15. Rasmussen J. Human error <strong>and</strong> the problem <strong>of</strong> causality in analysis <strong>of</strong> accidents. Philos<br />
Trans R Soc Lond B Biol Sci. 1990; 327: 449-60.<br />
16. Rasmussen J. Afterword. In: Bogner MS, editor. Human error in medicine. Hillsdale, N.J.:<br />
L. Erlbaum Associates; 1994: 385-93.<br />
17. Rasmussen J, Goodstein LP, Andersen HB, Olsen SE. Tasks, errors, <strong>and</strong> mental models : a<br />
festschrift to celebrate the 60th birthday <strong>of</strong> Pr<strong>of</strong>essor Jens Rasmussen. London ; New York<br />
:: Taylor & Francis,; 1988.<br />
18. Perrow C, Langton J. The limits <strong>of</strong> safety: the enhancement <strong>of</strong> a theory <strong>of</strong> accidents.<br />
Journal <strong>of</strong> Contigency Management. 1994; 2: 212-20.<br />
19. Perrow C. Normal accidents : Living with High-Risk Technologies. With a New Afterword<br />
<strong>and</strong> a Postscript on the Y2K Problem. Princeton, NJ:: Princeton University Press; 1999.<br />
20. Sagan SD. The Limits <strong>of</strong> <strong>Safety</strong>: Organizations, Accidents <strong>and</strong> Nuclear Weapons.<br />
Princeton, N.J.: Princeton University Press; 1993.<br />
455
21. Vaughn D. The Challenger Launch Decision : Risky Technology, Culture, <strong>and</strong> Deviance at<br />
Nasa. Chicago, Ill: Univ <strong>of</strong> Chicago Press; 1997.<br />
22. Sagan SD. The Challenger Launch Decision: Risky Technology, Culture, <strong>and</strong> Deviance at<br />
NASA. Administrative Science Quarterly. 1997; 42: 401-05.<br />
23. Roberts KH. The Challenger Launch Decision: Risky Technology, Culture, <strong>and</strong> Deviance<br />
at NASA. Administrative Science Quarterly. 1997; 42: 405-10.<br />
24. Weick KE. The Challenger Launch Decision: Risky Technology, Culture, <strong>and</strong> Deviance at<br />
NASA. Administrative Science Quarterly. 1997; 42: 395-401.<br />
25. Gaba DM. Structural <strong>and</strong> organizational issues in patient safety: a comparison <strong>of</strong> health<br />
care to other high-hazard industries. California Management Review. 2000; 43: 1-20.<br />
26. O'Brien JL, Shortell SM, Hughes EFX, Foster RW, Carman JM, Boerstler H, O'Connor EJ.<br />
An Integrative Model for Organization-wide Quality Improvement: Lessons from the Field.<br />
Quality Management in Health Care. 1995, 3:19-30.<br />
27. Weeks WB, Bagian JP. Developing a culture <strong>of</strong> safety in the Veterans Health<br />
Administration. Eff Clin Pract. 2000; 3: 270-6.<br />
28. Krause TR, Seymour KJ, Sloat KCM. Long-term evaluation <strong>of</strong> a behavior-based method<br />
for improving safety performance: a meta-analysis <strong>of</strong> 73 interrupted time-series<br />
replications. <strong>Safety</strong> Science. 1999; 32: 1-18.<br />
29. Helmreich RL, Merritt AC. Culture at work in aviation <strong>and</strong> medicine : national,<br />
organizational, <strong>and</strong> pr<strong>of</strong>essional influences. Aldershot ; Brookfield, VT, USA :: Ashgate;<br />
1998.<br />
30. Peters TJ, Waterman RH. In search <strong>of</strong> excellence : lessons from America's best-run<br />
companies. New York :: Harper & Row; 1982.<br />
31. Deal TE, Kennedy AA. Corporate cultures : the rites <strong>and</strong> rituals <strong>of</strong> corporate life.<br />
Reading, Mass. :: Addison-Wesley Pub. Co.; 1982.<br />
32. Spath P. Does your facility have a 'patient-safe' climate? Hosp Peer Rev. 2000; 25: 80-2.<br />
33. Baskin, Susan T (Osborn) <strong>and</strong> Shortell SM. Total Quality Management: needed research on<br />
the structural <strong>and</strong> cultural dimensions <strong>of</strong> quality improvement in health care organizations.<br />
The Journal <strong>of</strong> Health Administration Education. 1995; 13:143-154.<br />
34. Helmreich RL, Foushee HC, Benson R, Russini W. Cockpit resource management:<br />
exploring the attitude-performance linkage. Aviat Space Environ Med. 1986; 57: 1198-200.<br />
35. Cooper MD. Towards a model <strong>of</strong> safety culture. <strong>Safety</strong> Science. 2000; 36: 111-36.<br />
36. Geller ES. Ten Leadership qualities for a total safety culture. Pr<strong>of</strong>essional <strong>Safety</strong>. 2000; 45:<br />
30-32.<br />
37. Spath P. <strong>Patient</strong> safety improvement guidebook. Forest Grove, OR: Brown-Spath &<br />
Associates; 2000.<br />
38. Shortell SM, Jones RH, Rademaker AW, Gillies RR, Dranove DS, Hughes EF, et al.<br />
Assessing the impact <strong>of</strong> total quality management <strong>and</strong> organizational culture on multiple<br />
outcomes <strong>of</strong> care for coronary artery bypass graft surgery patients. Med Care. 2000; 38:<br />
207-17.<br />
39. Klein RL, Bigley GA, Roberts KH. Organizational culture in high reliability organizations:<br />
An extension. Human Relations. 1995; 48: 771-93.<br />
40. Billings CE, Reynard WD. Human factors in aircraft incidents: results <strong>of</strong> a 7-year study.<br />
Aviat Space Environ Med. 1984; 55: 960-5.<br />
41. Billings CE. Some hopes <strong>and</strong> concerns regarding medical event-reporting systems: lessons<br />
from the NASA Aviation <strong>Safety</strong> Reporting System. Arch Pathol Lab Med. 1998; 122: 214-<br />
5.<br />
456
42. Helmreich RL. On error management: lessons from aviation. BMJ. 2000; 320: 781-85.<br />
43. Barach P, Small SD. Reporting <strong>and</strong> preventing medical mishaps: lessons from non-medical<br />
near miss reporting systems. BMJ. 2000; 320: 759-63.<br />
44. Umiker W. Organizational culture: the role <strong>of</strong> management <strong>and</strong> supervisors. Health Care<br />
Superv. 1999; 17: 22-7.<br />
45. Glavin MPV, Chilingerian JA. Hospital care production <strong>and</strong> medical errors: organizational<br />
responses to improve care. Current Top Management. 1998; 3: 193-215.<br />
46. Manasee H, Jr. Close to home. Your own staff can solve many problems with your<br />
hospital's medication system. Hosp Health Netw. 2001; 75: 82.<br />
47. Findlay SE National Coalition on Health Care <strong>and</strong> the Institute for Healthcare<br />
Improvement. Reducing medical errors <strong>and</strong> improving patient safety. Available at:<br />
http://www.ihi.org/resources/act/medical_errors.pdf. Accessed June 26, 2001.<br />
48. Kohn L, Corrigan J, Donaldson M, editors. To Err Is Human: Building a Safer Health<br />
System. Washington, DC: Committee on Quality <strong>of</strong> Health Care in America, Institute <strong>of</strong><br />
Medicine. National Academy Press; 2000.<br />
457
Section G. Systems Issues <strong>and</strong> Human Factors<br />
Chapter 41. Human Factors <strong>and</strong> <strong>Medical</strong> Devices<br />
Subchapter 41.1. The Use <strong>of</strong> Human Factors in Reducing Device-related<br />
<strong>Medical</strong> Errors<br />
Subchapter 41.2. Refining the Performance <strong>of</strong> <strong>Medical</strong> Device Alarms<br />
Subchapter 41.3. Equipment Checklists in Anesthesia<br />
Chapter 42. Information Transfer<br />
Subchapter 42.1. Information Transfer Between Inpatient <strong>and</strong> Outpatient<br />
Pharmacies<br />
Subchapter 42.2. Sign-Out Systems for Cross-Coverage<br />
Subchapter 42.3. Discharge Summaries <strong>and</strong> Follow-up<br />
Subchapter 42.4. Notifying <strong>Patient</strong>s <strong>of</strong> Abnormal Results<br />
Chapter 43. Prevention <strong>of</strong> Misidentifications<br />
Subchapter 43.1. Bar Coding<br />
Subchapter 43.2. Strategies to Avoid Wrong-Site Surgery<br />
Chapter 44. Crew Resource Management <strong>and</strong> its Applications in Medicine<br />
Chapter 45. Simulator-Based Training <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong><br />
Chapter 46. Fatigue, Sleepiness, <strong>and</strong> <strong>Medical</strong> Errors<br />
Chapter 47. <strong>Safety</strong> During Transportation <strong>of</strong> <strong>Critical</strong>ly Ill <strong>Patient</strong>s<br />
Subchapter 47.1. Interhospital Transport<br />
Subchapter 47.2. Intrahospital Transport<br />
458
Chapter 41. Human Factors <strong>and</strong> <strong>Medical</strong> Devices<br />
Harvey J. Murff, MD<br />
Harvard <strong>Medical</strong> School<br />
John W. Gosbee, MD, MS<br />
Deparment <strong>of</strong> Veterans Affairs National Center for <strong>Patient</strong> <strong>Safety</strong><br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Introduction<br />
Human factors engineering (HFE), also known as usability engineering or ergonomics, is<br />
the study <strong>of</strong> how humans interact with machines <strong>and</strong> complex systems. 1 Through the merging <strong>of</strong><br />
cognitive psychology, engineering <strong>and</strong> other disciplines, human factors researchers have detailed<br />
numerous principles concerning device <strong>and</strong> s<strong>of</strong>tware program designs that allow for optimal<br />
usage. 2 When these principles are violated, improper use <strong>of</strong> a machine is more likely to result. 3,4<br />
Specific examples have been detailed during observations <strong>of</strong> user errors with electronic infusion<br />
devices. 5<br />
<strong>Medical</strong> device misuse is an important cause <strong>of</strong> medical error, 6,7 <strong>and</strong> therefore,<br />
incorporating human factors methodology into the design <strong>of</strong> medical devices has assumed an<br />
important role in ensuring patient safety. 2,8 This chapter first describes the use <strong>of</strong> HFE principles<br />
as a safety practice in the design <strong>of</strong> medical devices <strong>and</strong> their evaluation both prior to <strong>and</strong> after<br />
institutional purchase. Next, medical device alarms <strong>and</strong> the contribution <strong>of</strong> HFE to alarm<br />
improvements will be evaluated. Finally, the chapter reviews the use <strong>of</strong> preoperative checklist<br />
procedures to reduce anesthesia device failures (see also Chapter 23).<br />
Subchapter 41.1. The Use <strong>of</strong> Human Factors in Reducing Device-related <strong>Medical</strong> Errors<br />
Background<br />
Human factors engineering is a powerful component in the design <strong>of</strong> usable, safe medical<br />
devices. 8 HFE principles can be incorporated as safety practices that occur at various points<br />
during device development <strong>and</strong> usage. Industry can use HFE principles at multiple times in the<br />
design <strong>and</strong> developmental cycle <strong>of</strong> medical devices <strong>and</strong> s<strong>of</strong>tware packages. 3 Health care<br />
institutions can consider results <strong>of</strong> HFE evaluations when deciding which products to purchase.<br />
Finally, HFE principles can also be incorporated into the ongoing evaluation <strong>of</strong> devices that have<br />
already been purchased <strong>and</strong> are in use. While these practices have high face validity, there has<br />
been little formal study <strong>of</strong> their effectiveness in reducing medical error. They are presented here<br />
because they may hold promise if scrutinized rigorously, <strong>and</strong> to familiarize readers with their<br />
potential to reduce medical error.<br />
Design <strong>and</strong> Developmental Phase<br />
Data collected by the United States Food <strong>and</strong> Drug Administration (FDA) in the late<br />
1980s demonstrated that almost half <strong>of</strong> all medical device recalls resulted from design flaws. 9 In<br />
1990, Congress passed the Safe <strong>Medical</strong> Devices Act, giving the FDA the ability to m<strong>and</strong>ate<br />
good manufacturing practices (GMP). These GMP involve design controls for manufacturers<br />
that help ensure the use <strong>of</strong> HFE within medical device design. 9 As described in the Good<br />
459
Manufacturing Practice Regulation, Design Control subsection (Title 21-Section 820.30), these<br />
include the use <strong>of</strong> iterative design <strong>and</strong> testing during the developmental phase. The Act requires<br />
that designs be “appropriate <strong>and</strong> address the intended use <strong>of</strong> the device, including the needs <strong>of</strong><br />
the user <strong>and</strong> patient.” 10 Multiple human factors techniques, such as user studies, prototype tests,<br />
<strong>and</strong> task/function analysis, are utilized in the development <strong>and</strong> design process.<br />
Manufacturers are required not only to use human factors principles to repeatedly test the<br />
product in all phases <strong>of</strong> design, but also to validate the ultimate device design. Validation entails<br />
testing the device, either in an actual clinical situation or a simulation, <strong>and</strong> documenting that the<br />
device conforms to the individual user’s needs. Thus, manufacturers are required to apply HFE<br />
methods through the multiple phases <strong>of</strong> device design <strong>and</strong> development cycles. 10<br />
Human factors engineering practices for medical device design <strong>and</strong> evaluation have been<br />
well described. In 1993 the Association for the Advancement <strong>of</strong> <strong>Medical</strong> Instrumentation <strong>and</strong> the<br />
American National St<strong>and</strong>ards Institute established guidelines for the incorporation <strong>of</strong> HFE<br />
principles into medical device design. 11 This comprehensive document helped direct attention to<br />
the problem <strong>of</strong> poor medical device design <strong>and</strong> helped establish the design st<strong>and</strong>ards necessary<br />
to ensure safe medical equipment <strong>and</strong> thus should st<strong>and</strong> as the safety benchmark for industry.<br />
Only limited data are available concerning the application <strong>of</strong> HFE principles to medical<br />
device design, <strong>and</strong> most are not published. Nonetheless, the application <strong>of</strong> human factors<br />
principles during a device’s design phase has been demonstrated to reduce user error. <strong>Patient</strong><br />
controlled analgesia (PCA) pumps are a case in point <strong>of</strong> how HFE principles in product design<br />
reduce user error. User errors associated with poor interface design have been described with<br />
PCA pumps. 12,13 Lin <strong>and</strong> colleagues investigated whether applying human factors engineering<br />
principles to the design <strong>of</strong> the user interface <strong>of</strong> a PCA pump could result in fewer dosage errors<br />
as well as less time spent programming the device. 14 Information on device usage was obtained<br />
through cognitive task analysis. This involved observing <strong>and</strong> interviewing nurses operating PCA<br />
pumps both in the laboratory setting <strong>and</strong> in the field. Utilizing this feedback, as well as other<br />
human factors design principles, a “new” PCA pump interface was designed. Twelve recovery<br />
room nurses were required to complete specific tasks with both the st<strong>and</strong>ard PCA user interface<br />
<strong>and</strong> the newly designed interface. There were 29 programming errors on the traditional interface<br />
<strong>and</strong> 13 on the redesigned interface (an error reduction <strong>of</strong> 55%, p
Device Evaluation Prior to Purchase<br />
Adhering to HFE principles during initial design stages <strong>of</strong> a medical device is essential.<br />
However, human factors analysis should also be incorporated into the institutional decision to<br />
acquire a new medical device or s<strong>of</strong>tware program. 3 Device purchasers should strongly consider<br />
institution-specific human factors testing. Usability testing at the institutional level establishes<br />
built-in redundancies to capture any design problems missed by manufacturers. Furthermore, the<br />
users <strong>and</strong> environments at individual institutions will differ, possibly in important ways, from the<br />
users <strong>and</strong> environments in which the device or program was initially designed <strong>and</strong> tested. It is<br />
important for an institution to be aware <strong>of</strong> who the intended users <strong>of</strong> the device or s<strong>of</strong>tware will<br />
be, as well as where <strong>and</strong> when they plan to use the device. The information for such evaluations<br />
may be obtained from vendors, from an in-house analysis, or from independent organizations.<br />
Vendors must be able to prove to the FDA that the user will be able to operate the<br />
medical device in the way in which it was intended. 10 As companies are required to collect<br />
human factors analysis data, it is important that institutions wishing to purchase a new medical<br />
device or s<strong>of</strong>tware receive <strong>and</strong> carefully review this information. Gosbee provides a list <strong>of</strong><br />
questions to ask a vendor before a purchase, which include: “How long does it take to learn to<br />
operate the system? How long does it take to complete typical set-up tasks? What are the types<br />
<strong>and</strong> frequency <strong>of</strong> errors that could happen, <strong>and</strong> the systems to thwart them?” 3<br />
It is also important to consider the environment in which a device will be used.<br />
Idiosyncratic features <strong>of</strong> the environment, such as excessive noise or poor lighting, <strong>and</strong><br />
differences in user skill or acuity due to fatigue or otherwise, may affect safety <strong>and</strong> the device’s<br />
in-house usability.<br />
Some institutions have developed in-house usability labs, in order to rigorously test any<br />
device before purchasing. The Mayo Clinic uses simulations to test the usability <strong>of</strong> medical<br />
s<strong>of</strong>tware before purchasing. 17 By carefully measuring user performance with the s<strong>of</strong>tware they<br />
are able to uncover latent errors in the design. The usability lab is also able to measure the time<br />
necessary to learn to use the new s<strong>of</strong>tware. This important information can help predict the<br />
device’s or s<strong>of</strong>tware’s influence on workflow as well as its predilection for operator misuse.<br />
Even without sophisticated usability laboratories, an institution can use basic human<br />
factors techniques to evaluate a product before purchase. 3 Powerful techniques such as cognitive<br />
walk-through can be easily utilized at any institution. This involves observing the end-users <strong>of</strong> a<br />
product interact with the product. As they attempt to use the device, they are instructed to “think<br />
out loud.” Careful observation <strong>of</strong> the user’s actions <strong>and</strong> comments can identify potential design<br />
flaws that might make it difficult to utilize the device or s<strong>of</strong>tware.<br />
Independent organizations are another potential source <strong>of</strong> information on device safety.<br />
Unfortunately, most independent sources do not make clear to what degree HFE principles were<br />
used in product evaluations, although they do provide some assessment <strong>of</strong> safety. One such<br />
organization is ECRI (formerly the Emergency Care Research Institute), a nonpr<strong>of</strong>it international<br />
health services research agency. Another is the Institute <strong>of</strong> Safe <strong>Medical</strong> <strong>Practices</strong> (ISMP). Both<br />
release newsletters <strong>and</strong> publications regarding product safety. By searching these <strong>and</strong> similar<br />
databases, institutions can gather additional information concerning product safety prior to<br />
purchasing a device. ERCI also publishes articles specifically geared to the institutions that<br />
might wish to purchase a medical device or s<strong>of</strong>tware.<br />
Regardless <strong>of</strong> the level <strong>of</strong> pre-procurement testing, some unsafe designs will not be<br />
detected until after the product is in use. 3 Therefore, it is important for institutions to<br />
continuously evaluate these products to ensure safety.<br />
461
Ongoing Device Evaluation<br />
Devices <strong>and</strong> s<strong>of</strong>tware at greatest risk for user error should be systematically evaluated.<br />
This is particularly important in areas where multiple devices are used with different interfaces,<br />
such as the operating room or the intensive care units. 3 Furthermore, areas where multiple<br />
medications are stored together should be scrutinized for potential latent errors within device or<br />
s<strong>of</strong>tware user interfaces prior to user errors occurring.<br />
Resources are available that can help direct an institution’s search. Through publications<br />
from the FDA, ECRI, ISMP <strong>and</strong> similar organizations, medical device problems identified at<br />
other institutions can be targeted. Thus an important safety practice may be using this published<br />
information to search for latent errors within already purchased medical devices <strong>and</strong> applying<br />
this information toward a directed product evaluation at the local institution.<br />
Another potential safety practice is to educate practitioners about HFE principles to<br />
increase awareness <strong>of</strong> medical device user error. 3 Several groups, including the American<br />
Nurses’ Credentialing Center <strong>and</strong> the American Society <strong>of</strong> Health-Systems Pharmacists,<br />
recommend incorporating HFE training within health care curricula as a means to reduce error. 18<br />
To create a culture <strong>of</strong> safety within medicine, practitioners must couple the ability to<br />
identify potential design weaknesses with a change in the prevailing culture <strong>of</strong> silence<br />
surrounding medical errors. Educational programs directed at health care providers in training<br />
should address both <strong>of</strong> these important concerns. Curricula for teaching medical student <strong>and</strong><br />
medical residents HFE principles have been described 18 <strong>and</strong> will likely be adopted at other<br />
institutions. Casarett <strong>and</strong> Helms caution that an unintended result <strong>of</strong> error curriculum 19 may be<br />
that residents become too willing to attribute an error to system causes. Their concern is that the<br />
resident will ignore any possible individual contribution to the adverse medical event <strong>and</strong> not<br />
learn from analyses <strong>of</strong> the event. This concern has been discounted by Gosbee, stating that any<br />
error-in-medicine curriculum should aim to “teach residents to see when errors are due to<br />
inadequate skills <strong>and</strong> knowledge versus when they are due to inherent cognitive limitations <strong>and</strong><br />
biases.” 18<br />
Subchapter 41.2. Refining the Performance <strong>of</strong> <strong>Medical</strong> Device Alarms<br />
Background<br />
Numerous aspects <strong>of</strong> patient care compete for providers’ attention <strong>and</strong> can reduce their<br />
vigilance in monitoring medical devices. Alarms can alert providers to urgent situations that<br />
might have been missed due to other distractions <strong>and</strong> have become a necessary part <strong>of</strong> patient<br />
monitoring. In a study looking at critical incidents within a neonatal intensive care unit, 10%<br />
were detected through alarms. 20<br />
However, fundamental flaws in the design <strong>of</strong> current alarm systems likely decrease their<br />
impact. 21 There are reports documenting some alarm failings in the medical literature, 22 but few<br />
data address interventions to improve alarm system effectiveness. For an alarm to be effective it<br />
requires that a medical problem trigger the alarm, that personnel identify the source <strong>and</strong> reason<br />
for the alarm, <strong>and</strong> that the medical problem be corrected prior to patient injury. This section<br />
reviews 2 aspects <strong>of</strong> alarm safety: (1) the use <strong>of</strong> HFE principles in the redesign <strong>of</strong> medical alarms<br />
to improve identification <strong>of</strong> the source <strong>and</strong> reason for alarm, <strong>and</strong> (2) practices in both device<br />
design <strong>and</strong> programming that may improve safety by decreasing false positive alarms.<br />
462
Identification <strong>of</strong> Alarm Source <strong>and</strong> Reason<br />
The recognition accuracy <strong>of</strong> alarms within the operating room is quite low. When<br />
presented with alarm sounds <strong>and</strong> asked to identify the source, anesthesiologists, operating room<br />
technicians, <strong>and</strong> operating room nurses correctly identify the device producing the alarm only 33<br />
to 53.8% <strong>of</strong> the time. 23-25 Furthermore, experiments suggest that humans have difficulty reliably<br />
recognizing more than 6 alarms at one time. 26 The sheer number <strong>of</strong> different medical devices<br />
with alarms can make it difficult to discern one alarm from another <strong>and</strong> studies within the human<br />
factors literature have documented the inability <strong>of</strong> medical providers to discern between high<br />
priority <strong>and</strong> low priority alarms. 27 While this is a known problem in operating rooms <strong>and</strong><br />
intensive care units, how well alarms are recognized in other settings has not been described.<br />
Some effort has been made to improve alarm systems through redesign. 28 One nonmedical<br />
study examined ways to improve the recognition <strong>of</strong> auditory alarms by comparing<br />
abstract alarm sounds with specially designed alarms using speech <strong>and</strong> auditory icons. 29 Other<br />
studies within the human factors literature have revealed certain acoustical properties that are<br />
more likely to result in a higher sense <strong>of</strong> perceived urgency by the operator.<br />
In a series <strong>of</strong> experiments, Edworthy required subjects to rank the level <strong>of</strong> urgency<br />
associated with different alarms. 30 The acoustical properties <strong>of</strong> the alarms were altered for the<br />
different subjects. Level <strong>of</strong> urgency was then correlated with a specific alarm sound. After<br />
ranking a set <strong>of</strong> acoustic parameters based on perceived urgency, the experimenters predicted<br />
what urgency ranking the alarm would receive <strong>and</strong> played the alarms for a new set <strong>of</strong> subjects.<br />
The correlation between the subjects’ urgency rating <strong>and</strong> the investigators’ predicted ratings was<br />
93% (p
cases, with subjects serving as their own controls. In one study, subjects were required to<br />
identify when changes in physiologic parameters occurred using different visual formats. 34<br />
Response time <strong>and</strong> accuracy to the simulated cases was compared among a histogram, polygon,<br />
<strong>and</strong> numerical display. Subject responses were more accurate with the histogram <strong>and</strong> polygon<br />
displays (p=0.01).<br />
In the other study, 20 anesthesiologists with an average working experience <strong>of</strong> 5 years<br />
were required to perform specific tasks on an anesthesia simulator 35 (see Chapter 45). The tasks<br />
consisted <strong>of</strong> returning a set <strong>of</strong> abnormal hemodynamic parameters to normal using intravenous<br />
medications. A specific time for the completion was determined <strong>and</strong> this time was compared<br />
among 3 different visual interfaces. Trial time was significantly shorter with the traditional<br />
display (p
Decreasing the Frequency <strong>of</strong> Alarms<br />
Poorly designed device alarms can create not only problems with alarm recognition but<br />
also frequent false positive alarms. Two observational studies found that from 72 to 75% <strong>of</strong><br />
alarms during routine general anesthesia did not require corrective action. 36,37 Another study<br />
showed that only 3% <strong>of</strong> all auditory alarms during routine anesthesia monitoring represented a<br />
patient risk. 38 Providers frequently must interrupt clinical tasks to silence these false positive<br />
alarms. More concerning is the fact that when alarms are unreliable, they tend to be ignored. 21,39<br />
This “cry-wolf” effect is a significant detriment to the optimal performance <strong>of</strong> alarm systems <strong>and</strong><br />
may result in dire consequences when “true alarms” are ignored.<br />
False alarms can be managed in two ways. Devices can be designed so that they identify<br />
<strong>and</strong> eliminate false alarms before triggering or users can manipulate alarm parameters to reduce<br />
false alarms. User manipulation can range from adjusting alarm thresholds 40 to even turning the<br />
alarms <strong>of</strong>f. 22 There are no data describing how <strong>of</strong>ten operators reset alarm parameters to reduce<br />
false positive rates.<br />
Some research has focused on the identification <strong>of</strong> alarm parameters that improve or<br />
optimize alarm accuracy (ie, to improve the ratio <strong>of</strong> true positives to false positives—the “signalto-noise”<br />
ratio). For example, Rheineck-Leyssius <strong>and</strong> Kalkman studied how altering an alarm<br />
parameter on a pulse oximeter would affect the incidence <strong>of</strong> hypoxemia. 40 Consecutive patients<br />
admitted to the recovery room <strong>of</strong> a regional hospital in the Netherl<strong>and</strong>s after general or regional<br />
anesthesia were r<strong>and</strong>omized to either a lower limit <strong>of</strong> SpO2 90% or SpO2 85%. The 2 groups<br />
were comparable at baseline. The outcomes measured were hypoxemia, defined by a pulse<br />
oximeter reading less than or equal to 90% or 85%. The authors were also required to judge if<br />
they believed a signal to be artifact versus a true positive. The authors were blinded as to which<br />
group the subject was r<strong>and</strong>omized to during artifact assessment <strong>and</strong> data analysis. The relative<br />
risk <strong>of</strong> having a hypoxic episode (Sp02≤85%) in the group with the lower alarm limit set at 85%<br />
(as compared with those with the lower alarm limit set at 90%) was 3.10 (95% CI: 1.32-7.28,<br />
p
motion-detection testing to the identified pulse. The “smart” pulse-oximeter only triggered one<br />
false positive (an alarm that did not coincide with hypoxia) <strong>and</strong> had a relative risk <strong>of</strong> 0.09 (95%<br />
CI: 0.02-0.48) for generating a false positive alarm when compared with conventional pulse<br />
oximetry with a 21-second delay.<br />
Comment<br />
Observational studies have suggested that current alarm systems could be improved, but<br />
future laboratory studies are needed to determine which properties <strong>of</strong> alarm systems are most<br />
effective to alert operators. These tests must be followed by field studies <strong>and</strong> ultimately with<br />
trials looking at actual patient outcomes to determine the best designs. Information concerning<br />
the cost <strong>and</strong> feasibility <strong>of</strong> implementing these changes should also be gathered.<br />
False positive alarms remain a significant problem. Few data exist on the incidence <strong>of</strong><br />
resetting alarm parameters or at what parameter values alarm accuracy is optimized. Advances in<br />
alarm technology aimed at reducing false positives appear a promising alternative to resetting<br />
parameters.<br />
Subchapter 41.3. Equipment Checklists in Anesthesia<br />
Events related to medical equipment can be divided into two categories, user-error <strong>and</strong><br />
equipment failure. 43 Health device inspection <strong>and</strong> preventive maintenance by biomedical or<br />
clinical engineering departments have high face validity as an important patient safety practice in<br />
reducing equipment failure.<br />
There are many calls in the engineering literature to st<strong>and</strong>ardize equipment<br />
maintenance. 44-46,46 St<strong>and</strong>ardization <strong>of</strong> protocols is believed to help make the processes more<br />
efficient <strong>and</strong> reduce errors. 47 However, it has been difficult to st<strong>and</strong>ardize equipment<br />
maintenance practices due to a lack <strong>of</strong> the appropriate units on which to base measurement. 46<br />
Some authorities have suggested outcomes based on engineering endpoints such as reliability<br />
<strong>and</strong> accuracy. 48 Others have tried to validate a set <strong>of</strong> maintenance outcome units based on cost or<br />
quality metrics. 44,45,49 Some engineers have suggested the incorporation <strong>of</strong> clinical endpoints into<br />
medical equipment assessment. 48,50 Notwithst<strong>and</strong>ing differing views as to measurement <strong>of</strong><br />
endpoints, experts uniformly believe that st<strong>and</strong>ardization <strong>of</strong> engineering endpoints is vital to<br />
ensure adequately inspected <strong>and</strong> maintained equipment. 46 No studies to date have developed a<br />
widely used st<strong>and</strong>ardized protocol for equipment maintenance for clinical engineering<br />
departments, largely because the lack <strong>of</strong> st<strong>and</strong>ardization <strong>of</strong> endpoints renders assessing the<br />
relative value <strong>of</strong> any particular maintenance protocol impossible. 44-46,48,50 Nonetheless,<br />
equipment failure does result in a small fraction <strong>of</strong> clinical events <strong>and</strong> thus is an important safety<br />
intervention. Hopefully, future studies will help delineate the most effective practices for<br />
equipment maintenance processes.<br />
Use <strong>of</strong> checklists is another practice that helps ensure equipment readiness, particularly<br />
for equipment that is needed in critical situations <strong>and</strong>/or where equipment failure may have dire<br />
consequences. For example, a nurse at the beginning <strong>of</strong> each shift may use a checklist to ensure<br />
the readiness <strong>of</strong> a hospital ward’s resuscitation cart (“crash cart”) should it be needed (eg, the<br />
defibrillator is plugged-in <strong>and</strong> charged, the back-up suction pump works, medication is not past<br />
its expiration date). Similarly, a perfusion technologist can use a checklist to ensure cardiac<br />
bypass circuit <strong>and</strong> back-up equipment are ready before surgery. Published studies on the<br />
effectiveness <strong>of</strong> equipment checklists largely relate to the use <strong>of</strong> preoperative checklists to<br />
prevent anesthesia equipment failures since, to date, studies on the effectiveness <strong>of</strong> equipment<br />
466
checklists in medicine have been limited to this area. 51-53,54 These studies are reviewed in<br />
Chapter 23.<br />
Final Comment to Chapter 41<br />
Human factors testing is yielding important data regarding safe <strong>and</strong> effective medical<br />
device <strong>and</strong> alarm designs that take into account the users’ cognitive limitations. Machines can be<br />
designed <strong>and</strong> redesigned that enhance patient safety, rather than compromise it.<br />
Currently, there are no widely accepted st<strong>and</strong>ards for equipment maintenance intervals<br />
<strong>and</strong> protocols. Maintenance endpoints that incorporate clinical events as one component <strong>of</strong> the<br />
endpoint have been suggested. Until a reliable <strong>and</strong> validated engineering endpoint metric is<br />
widely recognized it will remain difficult to investigate the most effective maintenance practices.<br />
Other than the pioneering work in anesthesiology, HFE has been underutilized in<br />
medicine. Hopefully, in the near future, more attention will be focused on integrating human<br />
factors engineering within all aspects <strong>of</strong> medical training <strong>and</strong> practice, which will help create a<br />
culture <strong>of</strong> safety.<br />
References<br />
1. H<strong>and</strong>book <strong>of</strong> Human Factors <strong>and</strong> Ergonomics. 2 ed. New York: John Wiley <strong>and</strong> Sons,<br />
INC, 1997.<br />
2. Sawyer D. Do It By Design: An Introduction to Human Factors in <strong>Medical</strong> Devices. 1997.<br />
FDA.<br />
3. Gosbee J, Lin L. The Role <strong>of</strong> Human Factors Engineering in <strong>Medical</strong> Device <strong>and</strong> <strong>Medical</strong><br />
Systems Errors. In: Vincent C, editor. Clinical Risk Management: Enhancing <strong>Patient</strong><br />
<strong>Safety</strong>. 2000.<br />
4. Wiklund M, Gardner-Bonneau D, Carstensen P, Weinger M. Global st<strong>and</strong>ardization <strong>of</strong><br />
human factors for medical devices <strong>and</strong> systems. In: Proceedings <strong>of</strong> the Human Factors <strong>and</strong><br />
Ergonomics Society 44th Annual Meeting. Santa Barbara, CA: Human Factors <strong>and</strong><br />
Ergonomics Society, 2000: 533-536.<br />
5. Cook RI, Woods DD, Howie MB, Horrow JC, Gaba DM. Case 2-1992. Unintentional<br />
delivery <strong>of</strong> vasoactive drugs with an electromechanical infusion device. J Cardiothorac<br />
Vasc Anesth. 1992; 6:238-244.<br />
6. Cooper JB, Newbower RS, Kitz RJ. An analysis <strong>of</strong> major errors <strong>and</strong> equipment failures in<br />
anesthesia management: considerations for prevention <strong>and</strong> detection. Anesthesiology. 1984;<br />
60:34-42.<br />
7. Ludbrook GL, Webb RK, Fox MA, Singleton RJ. The Australian Incident Monitoring<br />
Study. Problems before induction <strong>of</strong> anaesthesia: an analysis <strong>of</strong> 2000 incident reports.<br />
Anaesth Intensive Care. 1993; 21:593-595.<br />
8. Bogner MS. Designing medical devices to reduce the likelihood <strong>of</strong> error. Biomed Instrum<br />
Technol. 1999; 33:108-113.<br />
9. Human Factors Implications <strong>of</strong> the New GMP Rule.<br />
http://www.fda.gov/cdrh/humfac/hufacimp.html . 4-22-1998.<br />
10. <strong>Medical</strong> Devices; Current Good Manufacturing Practice (CGMP) Final Rule; Quality<br />
System Regulation. http://www.fda.gov/cdrh/humfac/frqsr.html 61[195], 52601-52662. 10-<br />
7-1996. Food <strong>and</strong> Drug Administration.<br />
11. American National St<strong>and</strong>ards Institute AftAoMI. Human factors engineering guidelines<br />
<strong>and</strong> preferred practices for the design <strong>of</strong> medical devices (ANSI/AAMI HE-48). Arlington,<br />
VA: Association for the Advancement <strong>of</strong> <strong>Medical</strong> Instrumentation, 1993.<br />
467
12. Brown SL, Bogner MS, Parmentier CM, Taylor JB. Human Error <strong>and</strong> <strong>Patient</strong>-Controlled<br />
Analgesia Pumps. Journal <strong>of</strong> Intravenous Nursing. 1997; 20:311-316.<br />
13. Lin L. Human Error in <strong>Patient</strong>-Controlled Analgesia: Incident Reports <strong>and</strong> Experimental<br />
Evaluation. In: Proceedings <strong>of</strong> the Human Factors <strong>and</strong> Ergonomics Society 42nd Annual<br />
Meeting. Santa Barbara, CA: Human Factors <strong>and</strong> Ergonomics Society, 1998: 1043-1047.<br />
14. Lin L, Isla R, Doniz K, Harkness H, Vicente KJ, Doyle DJ. Applying human factors to the<br />
design <strong>of</strong> medical equipment: patient- controlled analgesia. J Clin Monit Comput. 1998;<br />
14(4):253-263.<br />
15. Aucella A, Kirkham T, Barnhart S, Murphy L, LaConte K. Improving ultrasound systems<br />
by user-centered design. In: Proceedings <strong>of</strong> the Human Factors <strong>and</strong> Ergonomics Society.<br />
Santa Barbara, CA: Human Factors <strong>and</strong> Ergonomics Society, 1994.<br />
16. Sawyer D. <strong>Medical</strong> device requirements, human factors, <strong>and</strong> the food <strong>and</strong> drug<br />
administration. In: Proceedings <strong>of</strong> the Human Factors <strong>and</strong> Ergonomics Society 44th<br />
Annual Meeting. Santa Barbara: Human Factors <strong>and</strong> Ergonomics Society, 2000: 526-527.<br />
17. Claus P, Gibbons P, Kaihoi, BH, Mathiowetz M. Usability lab: A new tool for process<br />
anaylsis at the Mayo Clinic. In: HIMSS Proceedings. Chicago: Healthcare Information<br />
Management Systems Society, 1997: 149-159.<br />
18. Gosbee J. Human Factors Engineering is the Basis for a Practical Error-in-Medicine<br />
Curriculum. First Workshop on Human Error in Clinical Systems.<br />
http://www.dcs.gla.ac.uk/~johnson/papers/HECS_99/Gosbee.html . 1999.<br />
19. Casarett D, Helms C. Systems error versus physicians' errors: Finding the balance in<br />
medical education. Acad Med. 1999; 74:19-22.<br />
20. Frey B, Kehrer B, Losa M, Braun H, Berweger L, Micallef J et al. Comprehensive critical<br />
incident monitoring in a neonatal-pediatric intensive care unit: experience with the system<br />
approach. Intensive Care Med. 2000; 26(1):69-74.<br />
21. Woods D. The Alarm Problem <strong>and</strong> Direct Attention in Dynamic Fault Management.<br />
Ergonomics. 1995; 38(11):2371-2393.<br />
22. Samuels SI. An alarming problem. Anesthesiology. 1986; 64(1):128.<br />
23. Finley GA, Cohen AJ. Perceived urgency <strong>and</strong> the anaesthetist: responses to common<br />
operating room monitor alarms. Can J Anaesth. 1991; 38(8):958-964.<br />
24. Loeb RG, Jones BR, Leonard RA, Behrman K. Recognition accuracy <strong>of</strong> current operating<br />
room alarms. Anesth Analg. 1992; 75(4):499-505.<br />
25. Momtahan K, Hetu R, Tansley B. Audibility <strong>and</strong> identification <strong>of</strong> auditory alarms in the<br />
operating room <strong>and</strong> intensive care unit. Ergonomics. 1993; 36:1159-1176.<br />
26. Patterson R, Milroy R. Auditory warnings on civil aircraft: the learning <strong>and</strong> retention <strong>of</strong><br />
warnings. Final Contract Report 7D/S/0142. 1980. Cambridge, MRC Applied Psychology<br />
Unit.<br />
27. Simons D, Frederiks T, Tappel J. The evaluation <strong>of</strong> an auditory alarm for a new medical<br />
device. In: Proceedings <strong>of</strong> the Human Factors <strong>and</strong> Ergonomics Society 41st Annual<br />
Meeting. Santa Barbara: Human Factors <strong>and</strong> Ergonomics Society, 1997: 777-781.<br />
28. Edworthy J, Stanton N. A user-centred approach to the design <strong>and</strong> evaluation <strong>of</strong> auditory<br />
warning signals: 1. Methodology. Ergonomics. 1995; 38:2262-2280.<br />
29. Leung YK, Smith S, Parker S, Martin R. Learning <strong>and</strong> Retentiono <strong>of</strong> Auditory Warnings.<br />
www.icad.org/websiteV2.0/conferences/icad97/leung.pdf . 2001.<br />
30. Edworthy J, Loxley S, Dennis I. Improving auditory warning design: relationship between<br />
warning sound parameters <strong>and</strong> perceived urgency. Hum Factors. 1991; 33:205-231.<br />
468
31. Stanford L, McIntyre J, Hogan J. Audible alarm signals for anaesthesia monitoring<br />
equipment. Int J Clin Monit Comput. 1985; 1:251-256.<br />
32. Burt JL, Bartolome DS, Burdette DW, Comstock JR. A psychophysiological evaluation <strong>of</strong><br />
the perceived urgency <strong>of</strong> auditory warning signals. Ergonomics. 1995; 38:2327-2340.<br />
33. Loeb RG. A measure <strong>of</strong> intraoperative attention to monitor displays. Anesth Analg. 1993;<br />
76:337-341.<br />
34. Gurushanthaiah K, Weinger MB, Englund CE. Visual display format affects the ability <strong>of</strong><br />
anesthesiologists to detect acute physiologic changes. A laboratory study employing a<br />
clinical display simulator. Anesthesiology. 1995; 83:1184-1193.<br />
35. Jungk A, Thull B, Hoeft A, Rau G. Ergonomic Evaluation <strong>of</strong> an Ecological Interface <strong>and</strong> a<br />
Pr<strong>of</strong>ilogram Display for Hemodynamic Monitoring. J Clin Monit. 1999; 15:469-479.<br />
36. Kestin IG, Miller BR, Lockhart CH. Auditory alarms during anesthesia monitoring.<br />
Anesthesiology. 1988; 69:106-109.<br />
37. S<strong>and</strong>erson P, Seagull FJ. Cognitive Ergonomics <strong>of</strong> Information Technology in <strong>Critical</strong><br />
Care: Contexts <strong>and</strong> Modalities for Alarm Interpretation. Available at:<br />
http://www.detir.qld.gov.au/hs/ergo97/s<strong>and</strong>erso.pdf . Accessed June 20, 2001.<br />
38. Xiao Y, Mackenzie F, Seagull J, Jaberi M. Managing the monitors: an analysis <strong>of</strong> alarm<br />
silencing activities during an anesthetic procedure. In: Proceeding <strong>of</strong> the Human Factors<br />
<strong>and</strong> Ergonomics Society 44th Annual Meeting. Santa Barbara: Human Factors <strong>and</strong><br />
Ergonomics Society, 2000: 250-253.<br />
39. Bliss JP, Gilson RD, Deaton JE. Human probability matching behavior in response to<br />
alarms <strong>of</strong> varying reliability. Ergonomics. 1995; 38(11):2300-2312.<br />
40. Rheineck-Leyssius AT, Kalkman CJ. Influence <strong>of</strong> pulse oximeter lower alarm limit on the<br />
incidence <strong>of</strong> hypoxaemia in the recovery room. Br J Anaesth. 1997; 79(4):460-464.<br />
41. Makivirta A, Koski EM. Alarms-inducing variability in cardiac postoperative data <strong>and</strong> the<br />
effects <strong>of</strong> prealarm delay. J Clin Monit. 1994; 10:153-162.<br />
42. Rheineck-Leyssius AT, Kalkman CJ. Advanced pulse oximeter signal processing<br />
technology compared to simple averaging. II. Effect on frequency <strong>of</strong> alarms in the<br />
postanesthesia care unit. J Clin Anesth. 1999; 11:196-200.<br />
43. Hyman WA. Errors in the Use <strong>of</strong> <strong>Medical</strong> Equipment. In: Bogner MS, editor. Human Error<br />
in Medicine. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1994: 327-347.<br />
44. Cohen T, Bakuzonis C, Friedman SB, Roa RL. Benchmark indicators for medical<br />
equipment repair <strong>and</strong> maintenance. Biomed Instrum Technol. 1995; 29:308-321.<br />
45. Cohen T. Validating medical equipment repair <strong>and</strong> maintenance metrics: a progress report.<br />
Biomed Instrum Technol. 1997; 31:23-32.<br />
46. Shaffer MJ. Biomedical equipment: maintenance factors, units <strong>of</strong> measurement, <strong>and</strong><br />
st<strong>and</strong>ards. Med Instrum. 1985; 19:268-272.<br />
47. Leape LL. Error in medicine. JAMA. 1994; 272:1851-1857.<br />
48. James PJ. Equipment management risk rating system based on engineering endpoints.<br />
Biomed Instrum Technol. 1999; 33:115-120.<br />
49. Cohen T. Validating medical equipment repair <strong>and</strong> maintenance metrics, Part II: Results <strong>of</strong><br />
the 1997 survey. Biomed Instrum Technol. 1998; 32:136-144.<br />
50. Keil O, Widmann DE. Assessment <strong>of</strong> the impact <strong>of</strong> medical devices on the quality <strong>of</strong> care.<br />
QRB Qual Rev Bull. 1984; 10:278-280.<br />
51. Caplan RA, Vistica MF, Posner KL, Cheney FW. Adverse anesthetic outcomes arising<br />
from gas delivery equipment: a closed claims analysis. Anesthesiology. 1997; 87:741-748.<br />
469
52. Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study<br />
<strong>of</strong> human factors. Anesthesiology. 1978; 49:399-406.<br />
53. Cooper JB. Toward prevention <strong>of</strong> anesthetic mishaps. Int Anesthesiol Clin. 1984; 22:167-<br />
183.<br />
54. March MG, Crowley JJ. An evaluation <strong>of</strong> anesthesiologists' present checkout methods <strong>and</strong><br />
the validity <strong>of</strong> the FDA checklist. Anesthesiology. 1991; 75:724-729.<br />
470
Chapter 42. Information Transfer<br />
Harvey J, Murff, MD<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Introduction<br />
<strong>Patient</strong> safety can be compromised by discontinuities in care. Studies suggest that<br />
discontinuity results from poor information transfer 1 <strong>and</strong> faulty communication, 2 which in turn<br />
may cause avoidable adverse events. 3<br />
Improving information transfer <strong>and</strong> communications among health care providers is an<br />
important patient safety practice <strong>and</strong> has been strongly recommended as a means to improve<br />
patient care. 1,3-7 This chapter evaluates safety practices involving improvements <strong>of</strong> provider-toprovider<br />
information transfer. <strong>Practices</strong> for evaluation include transfer <strong>of</strong> information between<br />
inpatient <strong>and</strong> outpatient pharmacies (Subchapter 42.1), sign-out systems for medical housestaff<br />
(Subchapter 42.2), automatically generated electronic discharge summaries (Subchapter 42.3),<br />
<strong>and</strong> systems to improve patient notification <strong>of</strong> abnormal results (Subchapter 42.4).<br />
Subchapter 42.1. Information Transfer Between Inpatient <strong>and</strong> Outpatient Pharmacies<br />
Background<br />
Accurate <strong>and</strong> timely information transfer between community <strong>and</strong> acute care pharmacies<br />
is an important safety practice. <strong>Patient</strong>s admitted to the hospital could benefit from the hospital’s<br />
pharmacy obtaining better information concerning their medication allergies as well as prior<br />
therapeutic failures. 8 Furthermore, when patients transition from acute care to outpatient care,<br />
changes in medications that occurred during hospitalization may cause confusion for both<br />
patients <strong>and</strong> providers. In one study surveying patients one week after hospital discharge,<br />
patients’ knowledge <strong>of</strong> their drug indications were worse for medications introduced during their<br />
hospitalization than for those taken prior to hospitalization (OR 0.69, 95% CI: 0.53-0.89). 9<br />
Confusion <strong>and</strong> incomplete information may increase the risk <strong>of</strong> under- or<br />
overmedication, harmful drug interactions, <strong>and</strong> other problems. Existing literature suggests that<br />
pharmacist interventions may reduce potential adverse drug events <strong>and</strong> have a modest impact on<br />
patient morbidity <strong>and</strong> mortality. 10,11 However, these studies have used independent reviewers to<br />
judge the impact <strong>of</strong> the intervention <strong>and</strong> have not specifically measured adverse drug events or<br />
patient outcomes. Clinical pharmacists’ consultations prior to discharge might also improve<br />
patient medication compliance (see Chapter 7). 12<br />
Uncontrolled studies report that information-exchange programs between hospital <strong>and</strong><br />
community pharmacies are perceived as beneficial <strong>and</strong> may have a positive impact on patient<br />
outcomes. 13 Although not the primary outcome measured in their small (n=127) observational<br />
study, Dvorak et al did note that using a pharmacy-to-pharmacy referral form was effective in<br />
preventing 2 medication errors. Thus, practices that improve information transfer between<br />
hospital <strong>and</strong> community pharmacies may improve patient safety.<br />
Of the many potential methods for improving information transfer between hospitals <strong>and</strong><br />
outpatient pharmacies, controlled trials have been reported in the literature for only 2:<br />
pharmaceutical care plans cards 14 <strong>and</strong> patient information facsimiles between pharmacies. 8<br />
Although direct electronic communication <strong>of</strong> pharmacy data may be superior to these methods,<br />
471
no controlled studies are currently available regarding this practice <strong>and</strong> therefore it is not<br />
reviewed within this chapter.<br />
Practice Description<br />
In a study by Smith <strong>and</strong> colleagues, patients received a card prior to discharge listing<br />
their pharmaceutical care plan, which included medication doses, indications, schedules, side<br />
effects, information as to the importance <strong>of</strong> drug compliance, <strong>and</strong> how to obtain medication<br />
refills. 14 <strong>Patient</strong>s were instructed to give the card to their community pharmacist. In another<br />
study, the intervention consisted <strong>of</strong> pharmacy-to-pharmacy facsimile transmission at the time <strong>of</strong><br />
admission <strong>and</strong> discharge from the hospital. 8 In this study, when a patient was admitted to the<br />
hospital their community pharmacy transmitted patient demographic information, historical<br />
information concerning allergies <strong>and</strong> adverse drug reactions, current medications, refill history,<br />
pharmacist’s monitoring notes, communications with patient <strong>and</strong> physician, <strong>and</strong> a detailed<br />
medication history to the admitting hospital. After discharge the hospital pharmacy transmitted<br />
to the community pharmacy a list <strong>of</strong> any potential medication problems identified by the hospital<br />
pharmacist on admission, the patient’s daily monitoring log, the pharmacist’s discharge<br />
summary, <strong>and</strong> a discharge medication table. The medical records department <strong>of</strong> the hospital also<br />
transmitted the patient’s discharge summary <strong>and</strong> laboratory test results to the community<br />
pharmacy.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Medication problems can arise because patients are frequently discharged from the acute<br />
care hospital on medications different from their ambulatory regimen. 15 Elderly patients in<br />
particular are at risk after discharge. 16 In one study, hospital providers changed 53% <strong>of</strong> the drugs<br />
prescribed by the primary care providers. 15<br />
The extent to which these medication changes <strong>and</strong> lack <strong>of</strong> communication result in<br />
recently discharged patients failing to receive medications or appropriate monitoring for their<br />
drug therapy is unclear. In one study <strong>of</strong> elderly patients, 32% <strong>of</strong> medications prescribed at<br />
discharge were not being taken 2 days after discharge. 17 Another study found that 51% <strong>of</strong><br />
patients recently discharged from acute care hospitals had deviated from their prescribed<br />
regimen. 18 Of those that had deviated from the prescribed drug regimen, 70% did not underst<strong>and</strong><br />
the medication regimen. In a Scottish study, recently discharged, elderly patients were issued a<br />
5-day supply <strong>of</strong> their medication on discharge <strong>and</strong> visited in their homes after these 5 days had<br />
elapsed. 16 Twenty-seven percent <strong>of</strong> the patients had not received a new prescription ordered on<br />
discharge. Of the patients with new prescriptions issued, 19% received inaccurately labeled<br />
medications. Medications were considered mis-labeled when non-specific container labels, such<br />
as “take as directed” replaced the more specific labels given on discharge. Some authors have<br />
suggested that improving communication about medications prior to <strong>and</strong> just after hospital<br />
discharge might reduce these medication errors. 18,19<br />
Poor communication is not the only problem. 16 <strong>Patient</strong> factors influence whether a<br />
medication is ultimately picked up <strong>and</strong> taken as directed. Deviations from prescribed drug<br />
regimens are multifactorial <strong>and</strong> improving pharmacy-to-pharmacy communication is only one<br />
aspect <strong>of</strong> the overall problem.<br />
Opportunities for Impact<br />
Data from primary care providers reveal that 96% <strong>of</strong> the respondents would like<br />
information concerning hospital drug changes. 20 Ninety-four percent <strong>of</strong> community pharmacists<br />
472
surveyed also wished to be provided with information concerning hospital drug changes. 20 We<br />
were unable to identify data regarding what percentage <strong>of</strong> hospital pharmacies routinely transfer<br />
information on patients’ medication regimens when they are admitted to <strong>and</strong> discharged from<br />
acute care.<br />
Study Design <strong>and</strong> Outcomes<br />
Two controlled studies were identified in the literature (Table 42.1.1). Both were<br />
r<strong>and</strong>omized trials but neither was blinded. In Smith et al, patients received a written pharmacy<br />
care plan at discharge. Home visits were made 7 to 10 days later to assess compliance <strong>and</strong><br />
discrepancies in the medication that patients were taking versus those ordered at discharge<br />
(Level 2). 14 In the study by Kuehl et al, 8 patients were r<strong>and</strong>omly assigned to either usual care or<br />
to a bi-directional exchange <strong>of</strong> pharmacy information by facsimile between the ambulatory<br />
pharmacy <strong>and</strong> the admitting hospital, upon admission <strong>and</strong> discharge (Level 1). The outcomes<br />
were pharmacist interventions, such as changing medication doses or making allergy<br />
recommendations (Level 2). 8<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Smith et al’s small study (n=53) <strong>of</strong> a pharmacy care plan card found that in both groups<br />
patients were taking different medications than those ordered at discharge. The authors found<br />
that compliance with post-discharge medications was significantly better in the group that had<br />
received the information card (p
group <strong>and</strong> the loss to follow-up group. This large loss to follow-up could have significantly<br />
altered their findings.<br />
Potential for Harm<br />
None <strong>of</strong> the studies evaluating different hospital-community pharmacy communication<br />
processes detailed any adverse events. However, the degree <strong>of</strong> added workload <strong>and</strong> effects on<br />
current workflow must be taken into account.<br />
Costs <strong>and</strong> Implementation<br />
The 2 reviewed interventions for improving hospital to community pharmacy information<br />
transfer seemed simple <strong>and</strong> relatively inexpensive. 8,8,13,14,14 Although no formal cost analysis was<br />
performed, Dvorak et al described their method (use <strong>of</strong> a pharmacy-to-pharmacy referral form)<br />
as being “labor-intensive.” No records were kept <strong>of</strong> the time necessary to provide the referrals<br />
but the authors estimated the time commitment per patient to be 30 minutes. 13 It is also important<br />
to note that complete information transfers occurred for 75% <strong>of</strong> the subjects in Kuehl’s study,<br />
indicating that there is still room for improving the process. With improvements in information<br />
transfer technology, automated transfer <strong>of</strong> information from hospital to community pharmacy<br />
could have important patient safety benefits without excessively increasing providers’<br />
workloads.<br />
Comment<br />
Providing patients with data forms to convey transfer <strong>of</strong> information between hospital<br />
<strong>and</strong> ambulatory pharmacies has potential for reducing discontinuities resulting from inadequate<br />
medication information. Few studies have evaluated the effect <strong>of</strong> these interventions on patient<br />
outcomes, although any improvement in the transfer <strong>of</strong> this information would likely be well<br />
received by ambulatory providers. Future studies are necessary to determine if <strong>and</strong> how<br />
improved pharmacy-to-pharmacy communications reduce preventable adverse drug events <strong>and</strong><br />
improve patient outcomes. Identifying the most effective <strong>and</strong> least disruptive forms <strong>of</strong> interpharmacy<br />
communication remains an area for further investigation.<br />
474
Table 42.1.1. <strong>Practices</strong> to improve transfer <strong>of</strong> medication information<br />
Study Study Setting Intervention Study Design,<br />
Outcomes<br />
Smith,<br />
1997 14<br />
Kuehl,<br />
1998 8<br />
References<br />
53 patients (>65 yrs<br />
old) discharged to<br />
home who were<br />
likely to experience<br />
difficulties with their<br />
medications<br />
156 patients<br />
admitted to small<br />
mid-western<br />
community hospital<br />
Copies <strong>of</strong> medication<br />
doses, indications, side<br />
effects, importance <strong>of</strong><br />
compliance <strong>and</strong> refill<br />
information given to<br />
patients at discharge<br />
Pharmacy-to-pharmacy<br />
facsimile transmission o<br />
medication regimen at<br />
time <strong>of</strong> admission <strong>and</strong><br />
discharge from hospital<br />
475<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 2<br />
Results<br />
<strong>Patient</strong>s taking medication<br />
not prescribed at discharge:<br />
information card 75%, control<br />
96% (p
13. Dvorak SR, McCoy RA, Voss GD. Continuity <strong>of</strong> care from acute to ambulatory care<br />
setting. Am J Health Syst Pharm. 1998; 55:2500-2504.<br />
14. Smith L, McGowan L, Moss-Barclay C, Wheater J, Knass D, Chrystyn H. An investigation<br />
<strong>of</strong> hospital generated pharmaceutical care when patients are discharged home from<br />
hospital. Br J Clin Pharmacol. 1997; 44:163-165.<br />
15. Himmel W, Tabache M, Kochen MM. What happens to long-term medication when<br />
general practice patients are referred to hospital? Eur J Clin Pharmacol. 1996; 50:253-257.<br />
16. Burns JM, Sneddon I, Lovell M, McLean A, Martin BJ. Elderly patients <strong>and</strong> their<br />
medication: a post-discharge follow-up study. Age Ageing. 1992; 21:178-181.<br />
17. Beers M, Sliwkowski J, Brooks J. Compliance with medication orders among elderly after<br />
hospital discharge. Hosp Formul. 1992; 27:720-724.<br />
18. Parkin D, Henney C, Quirk J, Crooks J. Deviations from prescribed drug treatment after<br />
discharge from the hospital. BMJ. 1976; 2:686-688.<br />
19. Katz E, Nicod P, Brunner HR, Waeber B. Changes in treatment during <strong>and</strong> after<br />
hospitalization in patients taking drugs for cardiovascular disease. Cardiovascular Drugs<br />
<strong>and</strong> Therapy. 1996; 10:189-192.<br />
20. Munday A, Kelly B, Forrester JW, Timoney A, McGovern E. Do general practitioners <strong>and</strong><br />
community pharmacists want information on the reasons for drug therapy changes<br />
implemented by secondary care? Br J Gen Pract. 1997; 47:563-566.<br />
Subchapter 42.2. Sign-Out Systems for Cross-Coverage<br />
Background<br />
As physicians go <strong>of</strong>f duty, they provide information to a “cross-covering” physician who<br />
will care for patients in the interim. The process <strong>of</strong> information transfer, known as “sign-out,” is<br />
<strong>of</strong>ten informal <strong>and</strong> unstructured. Various methods are used, including h<strong>and</strong>written lists, PCbased<br />
word processing or spreadsheet programs, <strong>and</strong> personal digital assistants (PDAs), but little<br />
literature has assessed their effectiveness in assuring continuity <strong>of</strong> care for patients <strong>and</strong><br />
preventing medical errors. Although notes in the medical record <strong>of</strong>ten contain all the information<br />
needed to care for patients, cross-covering physicians make many decisions without the benefit<br />
<strong>of</strong> the patients’ charts. 1 Jelley found lack <strong>of</strong> consistency in the content <strong>of</strong> weekend sign-out lists<br />
in a community-based internal medicine inpatient program. 1 Lee <strong>and</strong> colleagues found that<br />
medical interns recorded information elements such as patient age, DNR status, <strong>and</strong> medications<br />
more <strong>of</strong>ten when a st<strong>and</strong>ardized sign-out card was used. 2 In this section, we review evidence <strong>of</strong> a<br />
computerized sign-out system to reduce medical errors during cross-coverage.<br />
Practice Description<br />
The proposed safety practice is a structured sign-out process in which patient information<br />
is provided for various st<strong>and</strong>ardized data fields. The computerized sign-out program described<br />
by Peterson <strong>and</strong> colleagues consisted <strong>of</strong> a summary <strong>of</strong> the patient’s medical status, a problem<br />
list, recent laboratory data, resuscitation status, allergies, <strong>and</strong> a “to do” list. 3 This information<br />
was accessible from any computer within the hospital <strong>and</strong> was accessed <strong>and</strong> maintained on a<br />
daily basis by housestaff physicians.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Discontinuities in provider care during hospitalization have been associated with an<br />
increased risk <strong>of</strong> adverse events. The Petersen et al study found the odds ratio for a preventable<br />
476
adverse medical event occurring during cross-coverage as opposed to regular provider coverage<br />
to be 6.1 (95% CI: 1.4-26.7). 4 In the surgical domain, the Joint Commission on Accreditation <strong>of</strong><br />
Healthcare Organizations (JCAHO) publication on wrong-site surgeries noted a number <strong>of</strong> cases<br />
involving last minute personnel changes. 5 It is possible that these rapid substitutions in the<br />
operating room with inadequate communication may have contributed to these adverse events.<br />
Opportunities for Impact<br />
The number <strong>of</strong> hospitals using either a computerized or paper-based sign-out process is<br />
unknown, but computerized sign-outs are probably unusual. One study found that 26% <strong>of</strong><br />
adverse events in a single institution occurred during cross-coverage. 4 These data suggest that a<br />
st<strong>and</strong>ardized sign-out procedure could have a significant impact on improving patient safety.<br />
Study Design <strong>and</strong> Outcomes<br />
We identified one study evaluating the effect <strong>of</strong> a st<strong>and</strong>ardized sign-out system on the<br />
occurrence <strong>of</strong> adverse events (Table 42.2.1). In this study, adverse medical events detected by<br />
self-report were compared before <strong>and</strong> after implementation <strong>of</strong> a computerized sign-out system<br />
(Level 3). 3 This study involved internal medicine housestaff <strong>and</strong> any management <strong>of</strong> a patient<br />
performed by an intern from a different team or a night-float resident was considered crosscoverage.<br />
During chart review, the investigators recorded whether the physician at the time <strong>of</strong><br />
the event was the patient’s regular physician or a cross-covering physician. Adverse medical<br />
events, defined as “an injury due to medical therapy that prolonged hospital stay or disability at<br />
discharge” 3 were the primary outcomes (Level 2).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
There were significantly fewer adverse events during the intervention period compared<br />
with the baseline period (2.38% vs. 3.94%, p
Potential for Harm<br />
The study reported no adverse events as a result <strong>of</strong> the sign-out system. As with other<br />
sign-out systems, particularly those that are computerized or Web-enabled, the issues <strong>of</strong> data<br />
security <strong>and</strong> protection <strong>of</strong> confidentiality must be addressed. 6<br />
Costs <strong>and</strong> Implementation<br />
Implementation <strong>of</strong> a computerized sign-out system like that described by Peterson et al<br />
would require information systems that allow extraction <strong>and</strong> aggregation <strong>of</strong> patient specific data<br />
(eg, laboratory <strong>and</strong> pharmacy) <strong>and</strong> financial support for programming. These resources may not<br />
be present at some institutions. Although housestaff responded favorably to the computerized<br />
system, physicians at other institutions or in other specialties may not be as willing to use this<br />
system.<br />
Comment<br />
One study has shown that an inpatient’s risk <strong>of</strong> preventable adverse events was less after<br />
implementation <strong>of</strong> a computerized sign-out process. The method appears appropriately suited for<br />
hospitals with cross-coverage arrangements similar to those described by Peterson et al,<br />
specifically in-house coverage by resident trainees. It will be important to know if similar<br />
systems are as effective <strong>and</strong> well received at other institutions, including those without trainees.<br />
Such systems would be difficult to implement in hospitals with limited information systems or<br />
where physicians outside the hospital provide coverage through paging systems. Although a<br />
computerized system has the advantage <strong>of</strong> being accessible from any location in the hospital <strong>and</strong><br />
may be able to automatically import important information, events attributable to faulty<br />
communication during cross-coverage could also be amenable to other strategies for<br />
st<strong>and</strong>ardizing the process (eg, sign-out cards, Web-based programs, PDAs). No evidence is<br />
available concerning the relative effectiveness <strong>of</strong> other st<strong>and</strong>ardized sign-out methods. Future<br />
research should address what data fields are most helpful to physicians providing cross-coverage<br />
in preventing adverse events <strong>and</strong> how different methods <strong>of</strong> st<strong>and</strong>ardized sign-out compare in<br />
effectiveness (eg, h<strong>and</strong>written cards vs. PDAs).<br />
478
Table 42.2.1. Computerized sign-out program<br />
Study Setting Study Design,<br />
Outcome<br />
8767 patients admitted to<br />
the medical service <strong>of</strong> a<br />
tertiary care teaching<br />
hospital in Boston 3<br />
Level 3,<br />
Level 2<br />
Results (95% Confidence Intervals)<br />
Odds ratio (OR) <strong>of</strong> preventable adverse events<br />
occurring during cross-coverage compared with<br />
care under regular physician:<br />
baseline, OR 5.2 (1.5-18.2)<br />
with intervention, OR 1.5 (0.2-9.0)<br />
References<br />
1. Jelley MJ. Tools <strong>of</strong> continuity: the content <strong>of</strong> inpatient check-out lists. J Gen Intern Med.<br />
1994;9:77.<br />
2. Lee LH, Levine JA, Schultz HJ. Utility <strong>of</strong> a st<strong>and</strong>ardized sign-out card for new medical<br />
interns. J Gen Intern Med. 1996;11:753-755.<br />
3. Petersen LA, Orav EJ, Teich JM, O'Neil AC, Brennan TA. Using a computerized sign-out<br />
program to improve continuity <strong>of</strong> inpatient care <strong>and</strong> prevent adverse events. Jt Comm J<br />
Qual Improv. 1998;24:77-87.<br />
4. Petersen LA, Brennan TA, O'Neil AC, Cook EF, Lee TH. Does housestaff discontinuity <strong>of</strong><br />
care increase the risk for preventable adverse events? Ann Intern Med. 1994;121:866-872.<br />
5. The Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel event alert.<br />
Lessons learned: wrong site surgery. 1998. Available at:<br />
http://www.jcaho.org/edu_pub/sealert/sea6.html. Accessed June 27, 2001.<br />
6. Balsbaugh TA. The family physician <strong>and</strong> h<strong>and</strong>held computers: a beginner's guide. The<br />
Internet Journal <strong>of</strong> Family Practice 1[2]. 2001.<br />
Subchapter 42.3. Discharge Summaries <strong>and</strong> Follow-up<br />
Background<br />
Discharge summaries are important tools for communicating pertinent patient<br />
information regarding hospitalizations to outpatient care providers. Yet their relatively<br />
unstructured, narrative format <strong>of</strong>ten invites inaccuracies. 1 In addition, there can be significant<br />
delays transmitting discharge summaries to patients’ health care providers. 2,3<br />
Prior studies have investigated processes to improve discharge summaries, such as<br />
st<strong>and</strong>ardizing their format 4-6 <strong>and</strong> instituting physician education programs. 7 This chapter focuses<br />
on the use <strong>of</strong> structured, database-generated discharge summaries to improve the quality <strong>of</strong> the<br />
information content communicated after patient discharge, as well as to reduce the time required<br />
for this information transfer. 8<br />
Practice description<br />
During the hospital course, physicians provide information corresponding to specific<br />
sections <strong>of</strong> the computerized discharge summary either on data collection forms which are<br />
manually entered into a database 8 or directly into a computer system. When the patient is<br />
discharged the database generates a structured discharge summary that can be sent to the<br />
patient’s outpatient providers.<br />
479
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
In one study examining the effectiveness <strong>of</strong> inpatient follow-up care, 9.7% <strong>of</strong> discharged<br />
patients experienced worsening <strong>of</strong> symptoms or functional capacity as a result <strong>of</strong> an inadequately<br />
managed discharge process. 2 Hospital discharge summaries are an important means <strong>of</strong><br />
communication between hospital <strong>and</strong> community physicians, but have several problems. First,<br />
community physicians do not always receive summaries for recently discharged patients. In one<br />
study only 34% <strong>of</strong> patients had a discharge summary sent to their outpatient care provider. 2<br />
Although no analysis was undertaken to determine if receiving a discharge summary had an<br />
effect on patients’ follow-up, another study demonstrated that patients may be less likely to be<br />
readmitted to the hospital if their primary care provider receives a discharge summary. 9<br />
As mentioned above (Subchapter 42.1), patients frequently have their medication<br />
regimen changed while admitted. 10 The majority <strong>of</strong> ambulatory providers would like to have<br />
information regarding these medication changes. 11 Improvement in information transfer from<br />
acute care to ambulatory care might reduce medication discrepancies; however, patient<br />
compliance will also heavily influence these factors.<br />
Opportunities for Impact<br />
We found no data describing how many hospitals currently use database-driven discharge<br />
summaries.<br />
Study Design <strong>and</strong> Outcomes<br />
Several studies were identified that evaluated electronically generated discharge<br />
summaries, but these were limited by a lack <strong>of</strong> r<strong>and</strong>omization or limited outcomes reporting. 12-14<br />
One r<strong>and</strong>omized controlled trial was identified that compared traditional dictated discharge<br />
summaries to summaries generated from a database (Table 42.3.1). 8 The primary outcome was<br />
the proportion <strong>of</strong> admissions with a discharge summary completed by 4 weeks after discharge<br />
(Level 3). 8 Overall quality <strong>of</strong> the discharge summaries was also assessed (Level 3) but patient<br />
level outcomes were not.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
<strong>Patient</strong>s r<strong>and</strong>omized to the database group were significantly more likely to have a<br />
discharge summary generated within 4 weeks <strong>of</strong> discharge than were patients r<strong>and</strong>omized to the<br />
dictation group (113/142 vs. 86/151, p
Costs <strong>and</strong> Implementation<br />
The direct <strong>and</strong> indirect costs <strong>of</strong> implementing <strong>and</strong> maintaining a system for databasegenerated<br />
discharge summaries have not been formally evaluated in the literature. Results <strong>of</strong> a<br />
mail survey <strong>of</strong> housestaff in the van Walraven study 8 suggest their intervention did not adversely<br />
affect the workflow <strong>of</strong> housestaff physicians. Housestaff significantly preferred (p
6. Rawal J, Barnett P, Lloyd BW. Use <strong>of</strong> structured letters to improve communication<br />
between hospital doctors <strong>and</strong> general practitioners. BMJ 1993; 307:1044.<br />
7. Flyer B, Rubenstein LZ, Robbins AS, Wiel<strong>and</strong> GD, Henry D, Cugalj N. An intervention to<br />
improve the hospital discharge summary. J Med Educ 1988; 63:407-409.<br />
8. van Walraven C, Laupacis A, Seth R, Wells G. Dictated versus database-generated<br />
discharge summaries: a r<strong>and</strong>omized clinical trial. CMAJ 1999; 160:319-326.<br />
9. van Walraven C, Seth R, Austin P, Laupacis A. The effect <strong>of</strong> discharge summaries on<br />
readmission to hospital. 2001. [unpublished work]<br />
10. Himmel W, Tabache M, Kochen MM. What happens to long-term medication when<br />
general practice patients are referred to hospital? Eur J Clin Pharmacol 1996; 50:253-257.<br />
11. Munday A, Kelly B, Forrester JW, Timoney A, McGovern E. Do general practitioners <strong>and</strong><br />
community pharmacists want information on the reasons for drug therapy changes<br />
implemented by secondary care? Br J Gen Pract 1997; 47:563-566.<br />
12. Lissauer T, Paterson CM, Simons A, Beard RW. Evaluation <strong>of</strong> computer generated<br />
neonatal discharge summaries. Arch Dis Child 1991; 66(4 (Spec No)):433-436.<br />
13. Llewelyn DE, Ewins DL, Horn J, Evans TG, McGregor AM. Computerised updating <strong>of</strong><br />
clinical summaries: new opportunities for clinical practice <strong>and</strong> research? BMJ 1988;<br />
297:1504-1506.<br />
14. Smith RP, Holzman GB. The application <strong>of</strong> a computer data base system to the generation<br />
<strong>of</strong> hospital discharge summaries. Obstet Gynecol 1989; 73(5 Pt 1):803-807.<br />
15. Maresh M, Beard RW, Combe D, Dawson AM, Gillmer MDG, Smith G et al. Selection <strong>of</strong><br />
an obstetric data base for a microcomputer <strong>and</strong> its use for on-line production <strong>of</strong> birth<br />
notification forms, discharge summaries <strong>and</strong> perinatal audit. Br J Obstet <strong>and</strong> Gyn 1983;<br />
90:227-231.<br />
Subchapter 42.4. Notifying <strong>Patient</strong>s <strong>of</strong> Abnormal Results<br />
Background<br />
One <strong>of</strong> the most distressing safety issues <strong>of</strong> the clinical encounter is the failure to followup<br />
on diagnostic tests, particularly when a patient is not notified <strong>of</strong> an abnormal result. The<br />
complexities <strong>of</strong> this problem are legion. Contact methods—whether by phone, mail, fax or email,<br />
<strong>and</strong> whether sent by the lab, clinic or individual clinician—vary widely in their reliability<br />
(with most being imperfect). In some instances patients are told that if they do not hear back<br />
regarding their test results it signifies normal results. Of course, not hearing may mean that the<br />
test was lost or that the contact method was faulty. Other issues arise when the content <strong>of</strong> the<br />
notification is not clear, either as to result or the recommended follow-up for re-testing or<br />
treatment options.<br />
This chapter evaluates safety practices aimed at improving patient notification <strong>of</strong><br />
abnormal results. Adequate medical <strong>and</strong>/or surgical care during follow-up is essential to<br />
reducing patient morbidity <strong>and</strong> mortality, but practices to address this are beyond the scope <strong>of</strong><br />
patient safety as defined in this Report. We have chosen the example <strong>of</strong> Pap smear results,<br />
although many <strong>of</strong> the issues should be transferable to other laboratory (eg, PSA level) <strong>and</strong><br />
radiologic (eg, mammogram) results.<br />
482
Practice Description<br />
Our search revealed only one study evaluating patient notification practices. 1 In this<br />
study, the patient’s mailing address was included on the Pap smear request form. Two weeks<br />
after the patient’s primary care provider received the results, the laboratory directly notified the<br />
patient by mail. The notification was by form letter advising the patient <strong>of</strong> her results <strong>and</strong><br />
providing advice on the recommended follow-up step: discuss results with the doctor, return in<br />
two years time, or make an appointment to see the doctor without delay.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Few data exist concerning physician follow-up <strong>and</strong> patient notification <strong>of</strong> abnormal<br />
results. In a survey <strong>of</strong> attending physicians <strong>and</strong> residents practicing at a large urban teaching<br />
hospital <strong>and</strong> 21 suburban primary care practices, virtually all respondents believed it was<br />
moderately or extremely important to notify patients <strong>of</strong> abnormal results, yet 36% <strong>of</strong> physicians<br />
did not always do so. 2 Among the most common reasons reported by physicians were<br />
forgetfulness <strong>and</strong> inability to reach patients. One large cross-sectional study examined physician<br />
documentation <strong>of</strong> notification to patients <strong>of</strong> abnormal mammograms, Pap smears, <strong>and</strong><br />
cholesterol tests. The results demonstrated that certain patient characteristics such as race,<br />
language, <strong>and</strong> education may be associated with a failure to transmit abnormal results to<br />
patients. 3<br />
An estimated 4600 American women died <strong>of</strong> cervical cancer in 2000. 4 There are no data<br />
regarding to what extent delays in notification result in worse patient outcomes, including<br />
mortality. One study evaluating processes to reduce non-adherence rates with the follow-up <strong>of</strong><br />
abnormal Pap smears noted that many women (the exact number was not presented) reported<br />
that they were never notified <strong>of</strong> their abnormal Pap smear result initially. 5<br />
Tracking systems for abnormal Pap smear results have been briefly mentioned in the<br />
context <strong>of</strong> studies evaluating interventions to improve overall follow-up, not patient<br />
notification. 5,6 However, the effectiveness <strong>of</strong> these tracking systems was not specifically<br />
evaluated so they are not reviewed here.<br />
Opportunities for Impact<br />
Compliance rates with follow-up medical care after abnormal Pap smears typically<br />
ranges from 50% to 70%. 5,7-9 It is unclear how <strong>of</strong>ten losses to follow-up resulted from failure to<br />
notify. There are no data indicating how many practice groups directly mail abnormal Pap smear<br />
results to patients. Even with successful patient notification practices, a corresponding reduction<br />
in morbidity <strong>and</strong> mortality may not occur because <strong>of</strong> the other barriers to adequate follow-up<br />
described in the literature. 5,10<br />
Study Design <strong>and</strong> Outcomes<br />
The study reviewed for this chapter was a r<strong>and</strong>omized control design (Level 1). 1 (Table<br />
42.4.1). Providers were r<strong>and</strong>omized into 2 groups. In the intervention group, the pre-cervical<br />
smear questionnaire form had been redesigned to allow the patient to request that results be<br />
mailed directly to her. The physicians in the intervention group determined which patients would<br />
be <strong>of</strong>fered direct notification. <strong>Patient</strong>s <strong>of</strong> physicians in the control group were notified <strong>of</strong> results<br />
using whatever protocol the provider typically used. The authors did not elaborate on the<br />
methods used to notify patients in the control group.<br />
483
The primary outcome was adherence with follow-up visits (Level 2), which was defined<br />
by searching the laboratory records for evidence <strong>of</strong> a follow-up Pap smear one year after<br />
notification. If a cervical smear result was not located through the laboratory database search<br />
then the patient’s provider was contacted. 1<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Significantly fewer women with cervical intraepithelial neoplasia (CIN) on Pap smear<br />
who were r<strong>and</strong>omized to the intervention group were lost to follow-up (0/52 vs. 9/39 in the<br />
control group, p
Table 42.4.1. R<strong>and</strong>omized controlled trial <strong>of</strong> direct notification <strong>of</strong> abnormal Pap smear<br />
results*<br />
Study Study Setting Outcomes Results (95% Confidence Interval)†<br />
Del Mar,<br />
1995 1<br />
311 women with abnormal<br />
Pap smears from 42<br />
general practices in<br />
Australia<br />
Level 2 <strong>Patient</strong>s with CIN lost to follow-up:<br />
Direct mail notification: 0 (0-0.07)<br />
Control: 0.23 (0.11-0.39)<br />
<strong>Patient</strong>s with atypia lost to follow-up<br />
Direct mail notification: 0.10<br />
Control: 0.13 (p=NS)<br />
* CIN indicates cervical intraepithelial neoplasia; NS, not statistically significant.<br />
† Proportion <strong>of</strong> patients lost to follow-up reported in intervention group vs. control group.<br />
References<br />
1. Del Mar CB, Wright RG. Notifying women <strong>of</strong> the results <strong>of</strong> their cervical smear tests by<br />
mail: does it result in a decreased loss to follow-up <strong>of</strong> abnormal smears? Aust J <strong>Public</strong><br />
Health 1995; 19:211-213.<br />
2. Bookhaker EA, Ward RE, Uman JE, McCarthy BD. <strong>Patient</strong> notification <strong>and</strong> follow-up <strong>of</strong><br />
abnormal test results: a physician survey. Arch Intern Med 1996; 156:327-331.<br />
3. Burstin HR, Puoplolo AL, Haas JS, Cook EF, Brennan TA. <strong>Patient</strong>s at risk for missed<br />
follow-up after abnormal tests [abstract]. J Gen Intern Med 13, 46A. 1998.<br />
4. American Cancer Society. Cancer Facts & Figures 2000. Atlanta, GA . 2000.<br />
5. Marcus AC, Kaplan CP, Crane LA, Berek JS, Bernstein G, Gunning JE et al. Reducing<br />
loss-to-follow-up among women with abnormal Pap smears. Results from a r<strong>and</strong>omized<br />
trial testing an intensive follow-up protocol <strong>and</strong> economic incentives. Med Care 1998;<br />
36:397-410.<br />
6. Paskett ED, White E, Carter WB, Chu J. Improving follow-up after an abnormal Pap<br />
smear: a r<strong>and</strong>omized controlled trial. Prev Med 1990; 19:630-641.<br />
7. Marcus AC, Crane LA, Kaplan CP, Reading AE, Savage E, Gunning J et al. Improving<br />
adherence to screening follow-up among women with abnormal Pap smears: results from a<br />
large clinic-based trial <strong>of</strong> three intervention strategies. Med Care 1992; 30:216-230.<br />
8. Michielutte R, Diseker RA, Young LD, May WJ. Noncompliance in screening follow-up<br />
among family planning clinic patients with cervical dysplasia. Prev Med 1985; 14:248-258.<br />
9. Singer A. The abnormal cervical smear. Br Med J (Clin Res Ed) 1986; 293:1551-1556.<br />
10. Stewart DE, Buchegger PM, Lickrish GM, Sierra S. The effect <strong>of</strong> educational brochures on<br />
follow-up compliance in women with abnormal Papanicolaou smears. Obstet Gynecol<br />
1994; 83:583-585.<br />
11. Stewart DE, Lickrish GM, Sierra S, Parkin H. The effect <strong>of</strong> educational brochures on<br />
knowledge <strong>and</strong> emotional distress in women with abnormal Papanicolaou smears. Obstet<br />
Gynecol 1993; 81:280-282.<br />
12. Paskett ED, Phillips KC, Miller ME. Improving compliance among women with abnormal<br />
Papanicolaou smears. Obstet Gynecol 1995; 86:353-359.<br />
485
Final Comment to Chapter 42<br />
Faulty information transfer causes discontinuities <strong>of</strong> care that may result in adverse<br />
events. However, interventions to improve information transfer have received relatively little<br />
attention in the medical literature. Unfortunately, numerous barriers impede the appropriate<br />
transfer <strong>of</strong> information between institutions <strong>and</strong> between patient <strong>and</strong> provider. Future<br />
technologies that allow for more seamless transfer <strong>of</strong> information may mitigate these gaps in<br />
patient care. Further evaluation is critical.<br />
486
Chapter 43. Prevention <strong>of</strong> Misidentifications<br />
Heidi Wald, MD<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Subchapter 43.1. Bar Coding<br />
Background<br />
Machine-readable automatic identification (ID) systems, including bar codes, magnetic<br />
stripes, optical character recognition <strong>and</strong> radi<strong>of</strong>requency labeling, have improved productivity<br />
<strong>and</strong> quality in diverse industries. Bar codes represent the oldest <strong>and</strong> most common <strong>of</strong> these<br />
machine-readable ID systems, 1,2 <strong>and</strong> are widely used in industrial manufacturing, shipping <strong>and</strong><br />
inventory tracking operations. Prior to bar coding, these processes would have involved<br />
keystroke entry <strong>of</strong> identification numbers, producing approximately one error in every 300<br />
entered characters. In contrast, bar coding produces misidentification errors at rates ranging from<br />
one character in 15,000 to one character in 36 trillion. 3<br />
The use <strong>of</strong> bar coding in health care was first described over 30 years ago in clinical<br />
laboratories <strong>and</strong> blood banks. 1,4 In 1984, Rappoport identified 3 areas for the use <strong>of</strong> automatic ID<br />
technology in lab medicine: patient identification, document identification, <strong>and</strong> specimen<br />
identification. 1 However, in a 1987 survey by the American Hospital Association, the use <strong>of</strong> bar<br />
codes was most widespread in materials management departments, rather than in clinical<br />
application. 5 Other areas in which hospitals employed bar codes at that time included the clinical<br />
laboratory, pharmacy, radiology, medical records <strong>and</strong> asset management. Despite the Health<br />
Industry Bar Code Council’s call for st<strong>and</strong>ardization in the mid-1980s, the implementation <strong>of</strong> bar<br />
code technology has been stymied by lack <strong>of</strong> industry st<strong>and</strong>ards <strong>and</strong> failed cooperation among all<br />
stakeholders. 2,6<br />
As bar coding has the potential to substantially increase productivity <strong>and</strong> accuracy, one<br />
would expect it to be applied to important patient safety practices. Unfortunately, the published<br />
literature contains very little evidence regarding health care applications. In this chapter we focus<br />
on 4 areas in which bar coding shows promise for improving patient safety: patient<br />
identification, medication dispensing <strong>and</strong> administration, specimen h<strong>and</strong>ling, <strong>and</strong> medical record<br />
keeping. 2,7-10<br />
Practice Description<br />
<strong>Patient</strong> Identification<br />
Machine-readable patient ID systems could replace conventional wrist-b<strong>and</strong>ing, <strong>and</strong><br />
might reduce patient ID errors ranging from specimen collection to medication <strong>and</strong> blood<br />
product administration. Attempts to create such machine-readable ID systems were reported in<br />
the blood banking literature as early as 1977. 11 Transfusion Medicine is particularly attuned to<br />
issues <strong>of</strong> patient identification, as specimen collection <strong>and</strong> blood product administration account<br />
for the majority <strong>of</strong> preventable transfusion errors. 12,13<br />
Most scenarios for patient identification involve the substitution or supplementation <strong>of</strong><br />
the traditional wristb<strong>and</strong> for one with a unique bar code patient identifier. All patient specimens,<br />
medications <strong>and</strong> released blood products then receive the patient’s unique bar code ID. No<br />
487
procedure or treatment can occur unless the patient’s ID is scanned with a portable scanner <strong>and</strong><br />
matched with a bar code generated by the doctor’s order. For example, a phlebotomist would<br />
carry the scanner, check the patient’s ID against a bar coded specimen label or collection list,<br />
<strong>and</strong> draw blood only in the event <strong>of</strong> a match. Similarly, for administration or treatment, the<br />
patient’s ID <strong>and</strong> the intended therapeutic would be scanned at the bedside with a portable reader.<br />
If a match exists, the transfusion or medication is allowed <strong>and</strong> the time <strong>and</strong> date are recorded <strong>and</strong><br />
even transmitted directly to the hospital computer system. The nurse’s bar code ID can also be<br />
scanned <strong>and</strong> a timed administration record can be created. If there is no match, an alarm is<br />
sounded, <strong>and</strong> the administration delayed until the problem is resolved. 9,14<br />
Other technological means to reduce error in patient identification have been examined in<br />
the transfusion literature. Several researchers explored the use <strong>of</strong> a system providing a<br />
mechanical barrier to transfusion through a series <strong>of</strong> locks. 15 This system appears to be<br />
cumbersome <strong>and</strong> easily circumvented. In addition, electronic blood banking, with point-<strong>of</strong>-care<br />
crossmatching <strong>and</strong> computer-controlled release <strong>of</strong> blood, has been examined for high volume<br />
transfusion areas such as the operating room or intensive care unit. 16 Due to major barriers to<br />
large-scale implementation, these practices will not be discussed further.<br />
Specimen h<strong>and</strong>ling<br />
Clinical laboratories have integrated bar codes in specimen h<strong>and</strong>ling with a great deal <strong>of</strong><br />
success. 3,8 Several authors have described the development <strong>of</strong> central laboratory information<br />
systems (LIS) that employ bar code technology. Collection list <strong>and</strong> label s<strong>of</strong>tware can be<br />
modified to integrate <strong>and</strong> produce bar coded information. Confirmation <strong>of</strong> labeling <strong>and</strong> patient<br />
ID occurs at the bedside. Specimen sorting <strong>and</strong> aliquoting in the laboratory can be shortened or<br />
eliminated by various setups. At Rush-Presbyterian Hospital in Chicago, a central receiving<br />
station rapidly sorts <strong>and</strong> aliquots bar coded samples as they move on a conveyor belt. 3 At the<br />
University <strong>of</strong> Kansas <strong>Medical</strong> Center, the collection tubes are also used for analysis, thus<br />
eliminating the need for aliquoting samples. Additionally, the computer sends the bar codecoordinated<br />
orders to each <strong>of</strong> the 2 chemistry analyzers. Because a sample can then be<br />
appropriately processed at either analyzer, the need for sorting samples has also been<br />
eliminated. 8 Clinical labs also employ bar code technology in harder-to-automate processes. For<br />
instance, the University <strong>of</strong> Utah uses bar codes to replace common keystrokes for text reporting<br />
<strong>of</strong> microbiology results—eg, a technician might use a bar code “pick list” to scan the single bar<br />
code that means “no growth for 24 hours” <strong>and</strong> eliminate the need to type this phrase. 17,18<br />
Medication dispensing <strong>and</strong> administration<br />
The use <strong>of</strong> bar codes is uniquely suited to the dispensing <strong>and</strong> administration stages <strong>of</strong> the<br />
medication process. 19 Bar coding may be used to simplify the patient cassette (the medicine tray<br />
for bedside delivery) filling <strong>and</strong> verification process. 20 For instance, a technician fills a cassette<br />
according to a computerized, bar coded medication schedule <strong>and</strong> completes a quick verification<br />
by scanning the label <strong>of</strong> each unit-dose that has been placed in it with a h<strong>and</strong>held scanner. The<br />
computer can generate an error message if an incorrect medication is entered. The administration<br />
<strong>of</strong> bar coded medications can also be tracked at the point-<strong>of</strong>-care using a portable scanner <strong>and</strong><br />
compared against the hospital computer’s medication orders. 9,21 Bar coded medication<br />
administration has the added capability <strong>of</strong> creating a record <strong>of</strong> the administration (ie, RN, date,<br />
time) <strong>and</strong> a bill. This type <strong>of</strong> system could be integrated with a patient identification system in an<br />
attempt to eliminate errors resulting in administration <strong>of</strong> medication to the wrong patient.<br />
488
<strong>Medical</strong> record keeping<br />
Radiology <strong>and</strong> medical records departments use bar code technology to track the location<br />
<strong>and</strong> status <strong>of</strong> studies <strong>and</strong> charts. 22-24 Even more creative use <strong>of</strong> bar coded information has been<br />
reported in the emergency medicine <strong>and</strong> pharmacy literatures. As with the applications in the<br />
microbiology lab, bar codes can be used to replace frequently used text for the creation <strong>of</strong><br />
medical records. “Pick lists” <strong>of</strong> bar codes with their text equivalents can be employed in<br />
circumstances requiring speed <strong>and</strong> accuracy. Several uses <strong>of</strong> bar coded scripts have been<br />
examined in resuscitation events, <strong>and</strong> in mass casualty recording. 10,25-27 The use <strong>of</strong> bar code<br />
“pick lists” for the documentation <strong>of</strong> pharmacists’ clinical activities has also been explored. 28-30<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Bar code technology may be used to address any number <strong>of</strong> patient safety issues in<br />
medicine. For this discussion, we will define the target safety problem as patient identification in<br />
general, using transfusion medicine as a specific example.<br />
<strong>Patient</strong> identification remains a challenge in hospitals because <strong>of</strong> the number <strong>of</strong> complex<br />
interventions that occur to patients ranging from meals to surgeries. These interventions occur in<br />
a variety <strong>of</strong> locations <strong>and</strong> are provided by large teams <strong>of</strong> staff who work in shifts. In addition,<br />
sick patients, or those who have a language barrier, are not always capable <strong>of</strong> responding to<br />
questions about their identity or treatment plans. Hospitals generally rely on st<strong>and</strong>ardized<br />
wristb<strong>and</strong>s containing the patient’s name <strong>and</strong> other identifying information such as medical<br />
record number or date <strong>of</strong> birth. Unfortunately, conventional wristb<strong>and</strong>s are not reliable sources<br />
<strong>of</strong> patient identification. A 1991 national sample <strong>of</strong> 712 hospitals estimated error rates for<br />
conventional patient identification wristb<strong>and</strong>s to be 5.5%. 31 In half <strong>of</strong> the errors, the patient’s<br />
wristb<strong>and</strong> was absent altogether. The error rates were significantly lower in hospitals where<br />
phlebotomists had responsibility for monitoring wristb<strong>and</strong> accuracy, as the phlebotomy staff<br />
would not perform routine lab work unless the b<strong>and</strong> was corrected. Other errors included more<br />
than one wristb<strong>and</strong> with conflicting data (18.3%); wristb<strong>and</strong>s with incomplete (17.5%),<br />
erroneous (8.6%), or illegible data (5.7%); <strong>and</strong> rarely, patients wearing wristb<strong>and</strong>s with another<br />
patient’s data (0.5%). As patient identification data are only as good as the information entered<br />
at registration, the use <strong>of</strong> bar coded ID data could not be expected to correct certain types <strong>of</strong><br />
errors such as a wristb<strong>and</strong> with incorrect data entered at admission, although it is potentially<br />
beneficial in eliminating other types <strong>of</strong> errors such as illegible data.<br />
Even when wristb<strong>and</strong>s are free <strong>of</strong> errors, protocols for patient identification (such as dual<br />
witness verification <strong>of</strong> identification for blood transfusion) are easily circumvented or performed<br />
incorrectly. 32 In an analysis <strong>of</strong> major transfusion errors reported to the FDA over a 10-year<br />
period from 1976-1985, Sazama found 10 patient deaths where the actual <strong>and</strong> intended patients<br />
shared the same last name, <strong>and</strong> 5 deaths where the 2 shared the same hospital room. 13<br />
Consequently, automatic patient identification systems have been proposed as a technological<br />
solution to remove human factors (Subchapter 41.1) from the patient identification process.<br />
Despite technical improvements in testing for blood group identification, fatal ABOincompatible<br />
transfusions in the United States continue to occur at a rate ranging from<br />
approximately 1:600,000 to 1:800,000, with as many as two dozen fatalities in the US<br />
annually. 12,13 Thus, the chance <strong>of</strong> a patient suffering a fatal transfusion reaction due to ABOincompatibility<br />
is roughly equivalent to the risk <strong>of</strong> acquiring HIV infection from a blood<br />
transfusion. 33,34 <strong>Patient</strong> misidentification represents the most common cause <strong>of</strong> ABOincompatible<br />
transfusion, accounting for 46-57% <strong>of</strong> these errors. 12,13 Since the rate <strong>of</strong> patient <strong>and</strong><br />
489
donor having blood group compatibility by chance is approximately 60%, it is estimated that the<br />
total number <strong>of</strong> ABO-incompatible transfusions is much higher than the rate <strong>of</strong> fatal errors. A<br />
study from New York State estimated that as many as one in 12,000 transfusions involve<br />
administration <strong>of</strong> a blood product intended for another patient or release <strong>of</strong> blood <strong>of</strong> an incorrect<br />
group. *12<br />
Opportunities for Impact<br />
The implementation <strong>of</strong> automatic patient identification may present a large opportunity<br />
to bring transfusion medicine <strong>and</strong> other hospital interventions closer to the goal <strong>of</strong> zero risk.<br />
According to a survey conducted by the American Society <strong>of</strong> Hospital-System Pharmacists,<br />
1.1% <strong>of</strong> responding hospitals use bar coding <strong>of</strong> drug products in conjunction with bar coding on<br />
the patient’s identification tag. 35<br />
Study Designs<br />
Multiple reports <strong>of</strong> the use <strong>of</strong> bar codes appear in the medical literature, but most <strong>of</strong> these<br />
relate to inventory management. Few authors examine bar codes for patient identification. Only<br />
one <strong>of</strong> these studies was a prospective evaluation, <strong>and</strong> in the study bar coded patient<br />
identification comprised only one small part <strong>of</strong> the intervention. 10 One observational study<br />
examined a bar code patient ID system for medication administration. 9 In the study, however,<br />
routine patient ID scanning was easily circumvented <strong>and</strong> the actual error rate was not provided.<br />
The remainder <strong>of</strong> the reports are descriptive in nature. 14,36,37 Therefore, error rates in automated<br />
patient identification could not be readily compared to usual practice.<br />
The use <strong>of</strong> bar coding in other clinical care applications (aside from inventory control)<br />
has been examined prospectively in trauma recording 27 <strong>and</strong> in documenting pharmacists’<br />
interventions. 28 One additional observational study examined bar coding in pharmacy<br />
dispensing. 20<br />
*This is based on a reported error rate <strong>of</strong> 1/19,000 transfusions. The authors estimate a 64%<br />
chance that a r<strong>and</strong>om transfusion to an unintended recipient would be compatible <strong>and</strong> assume<br />
100% reporting <strong>of</strong> incompatible erroneous transfusions<br />
490
Study Outcomes<br />
Some <strong>of</strong> the studies report error rates in transfusion or medication errors (Level 2), while<br />
others report related outcomes such as speed <strong>of</strong> data entry (Level 3), transcription errors (Level<br />
2) in a variety <strong>of</strong> experimental <strong>and</strong> clinical settings, <strong>and</strong> user satisfaction (Level 3).<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
A point-<strong>of</strong>-care information system for medication management was implemented at a<br />
tertiary care center in Colorado. 9 The system provided online patient <strong>and</strong> medication data that<br />
verified medication administration at bedside using h<strong>and</strong>-held scanners to record patient ID,<br />
nurse ID, <strong>and</strong> medication unit-dose bar codes. When intervention data were compared with<br />
historical controls, the pre-intervention medication error rate <strong>of</strong> 0.17% dropped to 0.05%,<br />
sustained over 3 years for an overall decrease <strong>of</strong> 71% (p value not reported). There was a 33%<br />
decrease in "wrong drug" errors, a 43% decrease in "wrong time" errors, <strong>and</strong> a 52% decrease in<br />
"omitted dose" errors. There was a 47% decrease in "transcription/order-entry" errors. There was<br />
no change in "wrong patient" errors or "wrong dosage" errors, perhaps because the component <strong>of</strong><br />
the multifaceted intervention most likely to mitigate these errors, the use <strong>of</strong> the scanners for<br />
patient ID, was easily <strong>and</strong> frequently circumvented. It is unclear if bedside scanning added an<br />
unwanted layer <strong>of</strong> work for the nurses, or if they were uncomfortable performing this task in<br />
front <strong>of</strong> patients <strong>and</strong> families. Computerized pharmacy <strong>and</strong> bar code tracking <strong>of</strong> medications led<br />
to qualitative improvements in documentation time, scheduling <strong>of</strong> administration, nursingpharmacy<br />
communications, <strong>and</strong> pharmacist drug monitoring. However, the contribution <strong>of</strong> bar<br />
coding to the decreased error rate is not distinguishable from that <strong>of</strong> the entire intervention. It<br />
also appears that the point-<strong>of</strong>-care patient ID portion <strong>of</strong> the intervention was easily bypassed, <strong>and</strong><br />
was therefore not adequately evaluated. 9<br />
On a large medical ward <strong>of</strong> a university hospital, the satellite pharmacy was reconfigured<br />
to implement bar code technology for drug dispensing. 20 The hospital bar coded all medications,<br />
patient medication cassettes, patient wristb<strong>and</strong>s, <strong>and</strong> employees. St<strong>and</strong>ard dispensing time was<br />
estimated at 8.24 seconds, while dispensing time for bar coded medications was 6.72 seconds (p<br />
value not reported). Accuracy <strong>of</strong> the st<strong>and</strong>ard cassette fill system was 99.6% (equivalent to one<br />
error in 250 doses), while the accuracy <strong>of</strong> bar coded cassette fill was reported to be 100% (based<br />
on 0 errors in about 20,000 doses). The pharmacists were freed to do other work as the burden <strong>of</strong><br />
dispensing was shifted to pharmacy technicians.<br />
As pharmacists have begun to use bar code technology for medication distribution, 2<br />
pharmacy groups described the use <strong>of</strong> bar codes in recording their clinical interventions. 28, 29 One<br />
group found bar code documentation to have lower overall error rates compared to manual<br />
documentation, while the marginal cost <strong>of</strong> implementing bar coding was less than $100<br />
(factoring in savings in labor costs related to manual entry). 28 These data were limited by the<br />
small number <strong>of</strong> operators who differed on their preferences for manual versus bar code<br />
recording.<br />
A prospective trial reviewed bar code technology in trauma recording. Experienced<br />
emergency room nurses found bar coded pick lists to be easy to use <strong>and</strong> produced fewer errors<br />
per record compared with h<strong>and</strong>writing (2.63±0.24 vs. 4.48±0.3, p
The limitations <strong>of</strong> these studies are numerous, including their small sample sizes <strong>and</strong> lack<br />
<strong>of</strong> generalizability. However they demonstrate that bar code technology is generally easy for<br />
operators to use, can be applied in a variety <strong>of</strong> creative ways, <strong>and</strong> produces easy-to-demonstrate<br />
gains in accuracy <strong>and</strong> efficiency.<br />
Potential for Harm<br />
There is no clear detriment to patient identification with a bar code system. However, as<br />
with the addition <strong>of</strong> any new technology, the possibility exists that the complexity <strong>of</strong> the<br />
information system, especially if it grows, could create more routes for potential failure. For<br />
instance, an error during patient registration might be perpetuated throughout the hospitalization<br />
by the information systems, <strong>and</strong> be more difficult to correct than with conventional systems. The<br />
system’s data are only as accurate as that entered by fallible humans.<br />
Costs <strong>and</strong> Implementation<br />
Significant barriers need to be overcome before the full potential <strong>of</strong> bar coding can be<br />
exploited in the clinical setting. The process <strong>of</strong> medication dispensing <strong>and</strong> administration<br />
highlight several examples <strong>of</strong> these barriers. First, pharmaceutical manufacturers have yet to<br />
adopt a universal bar code st<strong>and</strong>ard like the UPC system used in grocery store inventory. 38 As <strong>of</strong><br />
yet, there has been no regulatory m<strong>and</strong>ate by the FDA to serve as an incentive, although the<br />
American Society <strong>of</strong> Hospital-System Pharmacists recently urged the FDA to take action. 39<br />
Second, placing bar codes on unit-doses <strong>of</strong> medications <strong>of</strong>ten requires major changes in<br />
packaging (such as an increase <strong>of</strong> the package size to accommodate the bar coded label). Bar<br />
code tracking systems would have difficulty with unusual doses such as halved tablets. IV doses<br />
would still require bar code labeling in the pharmacy when prepared.<br />
At this point, implementation <strong>of</strong> bar coding requires a commitment on the part <strong>of</strong> the<br />
hospital to relabel every unit-dose <strong>of</strong> medication using a hospital st<strong>and</strong>ard. The hospital<br />
pharmacies that have implemented these systems have repackaged <strong>and</strong> relabeled many unitdoses<br />
at considerable cost. 9,20 One health system estimated costs at $119,516 annually, with the<br />
per dose costs <strong>of</strong> bar code labeling estimated at 2.73 cents. 20 The bottom line is that, at present,<br />
the costs <strong>of</strong> implementing bar coding in an entire pharmacy inventory are significant <strong>and</strong> the<br />
logistics complex.<br />
The use <strong>of</strong> bar coding in simpler clinical scenarios (ie, using a blood transfusion<br />
wristb<strong>and</strong> for patient identification) may be implemented with very modest outlay <strong>of</strong> resources<br />
(estimated to be less than 5 cents per wristb<strong>and</strong>). 36 The costs <strong>of</strong> the scanners themselves are<br />
moderate. A new model scanner, s<strong>of</strong>tware <strong>and</strong> recharger were priced at about $1100 in 1997. 28<br />
Comment<br />
Bar coding is a fast <strong>and</strong> accurate method <strong>of</strong> automated data capture that in experimental<br />
settings provides qualitative improvements in speed <strong>and</strong> accuracy <strong>of</strong> data entry. Bar code<br />
technology might be creatively applied to any number <strong>of</strong> data-driven processes in medicine. As<br />
the rapid <strong>and</strong> accurate transfer <strong>of</strong> data is paramount in health care, the thoughtful application <strong>of</strong><br />
appropriately piloted <strong>and</strong> evaluated bar code technology is likely to be well received <strong>and</strong><br />
deserves further investigation.<br />
492
References<br />
1. Rappoport A. A hospital patient <strong>and</strong> laboratory machine-readable identification system<br />
(MRIS) revisited. J Med Syst. 1984;8:133-156.<br />
2. Weilert M, Tilzer LL. Putting bar codes to work for improved patient care. Clin Lab Med.<br />
1991;11:227-238.<br />
3. Maffetone MA, Watt SW, Whisler KE. Automated sample h<strong>and</strong>ling in a clinical<br />
laboratory. Comput Healthc. 1988;9:48-50.<br />
4. Brodheim E, Ying W, Hirsch R. An evaluation <strong>of</strong> the codabar symbol in blood bank<br />
automation. Vox Sang. 1981;40: 3.<br />
5. Longe K. The status <strong>of</strong> bar codes in hospitals: a survey report. Hospital Technology Series.<br />
Chicago: American Hospital Association; 1989. 8.<br />
6. Garza D, Murdock S, Garcia L, Trujillo JM. Bar codes in the clinical laboratory. Clin Lab<br />
Sci. 1991;4:23-25.<br />
7. Zarowitz BJ, Petitta A, Rudis MI, Horst HM, Hyzy R. Bar code documentation <strong>of</strong><br />
pharmacotherapy services in intensive care units. Pharmacotherapy. 1996;16:261-266.<br />
8. Tilzer LL, Jones RW. Use <strong>of</strong> bar code labels on collection tubes for specimen management<br />
in the clinical laboratory. Arch Pathol Lab Med. 1988;112:1200-1202.<br />
9. Puckett F. Medication-management component <strong>of</strong> a point-<strong>of</strong>-care information system. Am<br />
J Health-Syst Pharm. 1995;52:1305-1309.<br />
10. Noordergraff G, Bouman J, Van Den Brink E, Van De Pompe C, Savelkoul T.<br />
Development <strong>of</strong> computer-assisted patient control for use in the hospital setting during<br />
mass casualty incidents. Am J Emerg Med. 1996;14:257-261.<br />
11. Sherer P, Chambers R, Taswell H, Altshuler C, Aster R, Covino K, et al. Automated donorrecipient<br />
identification systems as a means <strong>of</strong> reducing human error in blood transfusion.<br />
Transfusion. 1977;17:586-597.<br />
12. Linden J, Paul B, Dressler K. A report <strong>of</strong> 104 transfusion errors in New York State.<br />
Transfusion. 1992;32:601-606.<br />
13. Sazama K. Reports <strong>of</strong> 355 transfusion-associated deaths: 1976 through 1985. Transfusion.<br />
1990;30:583-590.<br />
14. Jensen N, Crosson J. An automated system for bedside verification <strong>of</strong> the match between<br />
patient identification <strong>and</strong> blood unit identification. Transfusion. 1996;36:216-221.<br />
15. Wenz B, Burns E. Improvement in transfusion safety using a new blood unit <strong>and</strong> patient<br />
identification system as part <strong>of</strong> safe transfusion practice. Transfusion. 1991;31:401-403.<br />
16. Cox C, Enno A, Deveridge S, Seldon M, Richards R, Martens V, et al. Remote electronic<br />
blood release system. Transfusion. 1997;37:960-964.<br />
17. Aller R, Freidman W. Rapid accuate entry <strong>of</strong> microbiology results. Arch Pathol Lab Med.<br />
1996;120:57-61.<br />
18. Willard K, Stanholtzer C. User interface reengineering: Innovative applications <strong>of</strong> bar<br />
coding in a clinical microbiology laboratory. Arch Pathol Lab Med. 1995;119:706-712.<br />
19. Bates D. Using information technology to reduce rates <strong>of</strong> medication errors in hospitals.<br />
BMJ. 2000;320:788-791.<br />
20. Meyer G, Br<strong>and</strong>ell R, Smith J, Milewski F, Brucker J, P, Coniglio M. Use <strong>of</strong> bar codes in<br />
inpatient drug distribution. Am J Hosp Pharm. 1991;48:953-966.<br />
21. Gebhart F. VA slashed drug erros via bar-coding. Drug Topics. 1999; 143: 44.<br />
22. Mudie D. Bar coding in a medical records department: a case study. Hosp Technol Ser.<br />
1988;7:1-15.<br />
493
23. Arenson RL, London JW. Comprehensive analysis <strong>of</strong> a Radiology Operations Management<br />
computer system. Radiology. 1979;133:355-362.<br />
24. Connelly A. Bar codes help control incomplete medical records. J Am Med Rec Assoc.<br />
1987;58:48-49.<br />
25. Burt TW, Bock HC. Computerized mega code recording. Ann Emerg Med. 1988;17:339-<br />
342.<br />
26. Bock HC. Field verification methodology using bar coding to record data. Ann Emerg Med.<br />
1993;22:75-8; discussion 78-79.<br />
27. Chua R, Cordell W, Ernsting K, Bock H, Nyhuis A. Accuracy <strong>of</strong> bar codes versus<br />
h<strong>and</strong>writing for recording trauma recuscitation events. Ann Emerg Med. 1993;22:1545-<br />
1550.<br />
28. Kanmaz T, Haupt B, Peterson A. Comparison <strong>of</strong> manual <strong>and</strong> bar-code systems for<br />
documenting pharmacists' interventions. Am J Health-Syst Pharm. 1997;54:1623-1626.<br />
29. Zarowitz B, Petitta A, Mlynarek M, Touchette M, Peters M, Long P, et al. Bar-code<br />
technology applied to drug use evaluation. Am J Hosp Pharm. 1993;50:953-959.<br />
30. Scott M, McElnay J, Burnett K. Using bar-code technology to capture clinical intervention<br />
data in a hospital with a st<strong>and</strong>-alone pharmacy computer system. Am J Health-Syst Pharm.<br />
1996;53:651-654.<br />
31. Renner S, Howanitz P, Bachner P. Wristb<strong>and</strong> identification error reporting in 712<br />
hospitals. Arch Pathol Lab Med. 1993;117:573-577.<br />
32. Feldman S, Roblin D. <strong>Medical</strong> accidents in hospital care: applications <strong>of</strong> failure analysis to<br />
hospital quality appraisal. Jt Comm J Qual Improv. 1997;23:567-580.<br />
33. Schreiber G, Busch M, Kleinman S, Korelitz J. The risk <strong>of</strong> transfusion-transmitted viral<br />
infections. The Retrovirus Epidemiology Donor Study. N Engl J Med. 1996;334:1685-<br />
1690.<br />
34. Lackritz J, Satten G, Alberle-Grasse J, al e. Estimated risk <strong>of</strong> transmission <strong>of</strong> the human<br />
inmmunodeficiency virus by screened blood in the United States. N Engl J Med.<br />
1995;333:1721-1725.<br />
35. American Society <strong>of</strong> Hospital-System Pharmacists. National survey <strong>of</strong> pharmacy practice<br />
in acute care settings: dispensing <strong>and</strong> administration - 1999. Am J Health-Syst Pharm.<br />
2000;57:1759-1775.<br />
36. Lau F, Wong R, Chui C, Ng E, Cheng G. Improvement in transfusion safety using a<br />
specially designed transfusion wristb<strong>and</strong>. Transfusion Medicine. 2000;10:121-124.<br />
37. Amber R, Everett V. Emergency department patient tracking: A cost-effective system<br />
using bar code technology. J Emerg Nurs. 1996;22:190-195.<br />
38. Davis NM. Initiative for reducing medication errors: the time is now. Am J Hosp Pharm.<br />
2000;57:1488.<br />
39. American Society <strong>of</strong> Hospital-System Pharmacists. ASHP policy recommendation. Am J<br />
Health Syst Pharm. 2001;58:603-606.<br />
Subchapter 43.2. Strategies to Avoid Wrong-Site Surgery<br />
Background<br />
Operating on the wrong site or body part represents a potentially devastating event for all<br />
parties involved. Cases <strong>of</strong> “wrong-site surgery” frequently attract considerable media attention 1-4<br />
<strong>and</strong> foment malpractice lawsuits. Claims for wrong-site orthopedic surgeries result in indemnity<br />
494
payments in 84% <strong>of</strong> cases, compared with only 30% <strong>of</strong> orthopedic claims overall. 5,6 Although<br />
orthopedics represents the largest source <strong>of</strong> legal claims, the Physician’s Insurance Association<br />
<strong>of</strong> America (PIAA) has h<strong>and</strong>led wrong-site surgery litigation from the entire range <strong>of</strong> surgical<br />
specialties <strong>and</strong> subspecialties. 6 Common factors identified in wrong-site surgery include the<br />
involvement <strong>of</strong> multiple surgeons on a case, the performance <strong>of</strong> multiple procedures during a<br />
single trip to the operating room, unusual time constraints, <strong>and</strong> unusual anatomy or patient<br />
characteristics, such as physical deformity or morbid obesity. 7,8<br />
Based upon careful review <strong>of</strong> 43 cases reported through its Sentinel Event Policy 9 over a<br />
3-year period, the Joint Commission on the Accreditation <strong>of</strong> Healthcare Organizations (JCAHO)<br />
issued the following recommendations for avoiding wrong-site surgery: 7,8<br />
• Mark the operative site <strong>and</strong> involve the patient in this process<br />
• Require oral verification <strong>of</strong> the correct site in the operating room by each member <strong>of</strong><br />
the surgical team<br />
• Follow a verification checklist that includes all documents <strong>and</strong> medical records<br />
referencing the intended operative procedure <strong>and</strong> site<br />
• Directly involve the operating surgeon in the informed consent process<br />
• Engage in ongoing monitoring to ensure verification procedures are followed<br />
Among these recommendations, marking the operative site has received the most<br />
attention <strong>and</strong> is the focus <strong>of</strong> this chapter.<br />
Practice Description<br />
In 1998, the American Academy <strong>of</strong> Orthopaedic Surgeons endorsed a program <strong>of</strong><br />
preoperative surgical site identification called “Sign your Site,” modeled on the “Operate<br />
through your initials” campaign instituted by the Canadian Orthopaedic Association from 1994-<br />
96. 5,10 Both organizations recommend that the operating surgeon initial the intended operative<br />
site, using a permanent marker, during a scheduled preoperative visit. For spinal surgery, the<br />
recommendations additionally endorse the use <strong>of</strong> intra-operative x-rays for localization <strong>of</strong> the<br />
pathologic spinal level before proceeding with the procedure. Many surgeons already employ<br />
their own techniques for surgical site identification such as marking an “X” on the operative site<br />
or marking “No” on the wrong limb. 5,11,12 While these practices are commendable, they have<br />
theoretical drawbacks including lack <strong>of</strong> st<strong>and</strong>ardization across operating rooms <strong>and</strong> institutions.<br />
These alternative strategies will not be evaluated further in this chapter.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
From January 1995 to March 2001, JCAHO reviewed voluntary reports <strong>of</strong> 1152 “sentinel<br />
events.” Wrong-site surgery accounted for 114 (9.9%) <strong>of</strong> these reports <strong>and</strong> included procedures<br />
in neurosurgery, urology, orthopedics, <strong>and</strong> vascular surgery. 13 Despite the high pr<strong>of</strong>ile <strong>of</strong><br />
JCAHO’s Sentinel Event Policy, 9 under-reporting by health care organizations almost certainly<br />
affects these statistics. Only 66% <strong>of</strong> the 1152 total events were “self-reported” by the institutions<br />
involved. The remainder came from patient complaints, media stories <strong>and</strong> other sources. 13 In<br />
fact, using a m<strong>and</strong>atory reporting system, the New York State Department <strong>of</strong> Health received 46<br />
reports <strong>of</strong> wrong-site surgery from April 1, 1998 through March 31, 2000 4 (F. Smith, personal<br />
communication, May 2001), compared with the 114 cases JCAHO received nationally over a<br />
495
period 3 times longer. 13 This suggests that voluntary incident reporting may underestimate the<br />
true incidence by a factor <strong>of</strong> 20 or greater. 14 †<br />
The PIAA reviewed claims data from 22 malpractice carriers representing 110,000<br />
physicians from 1985 to 1995. 5 These claims included 331 cases <strong>of</strong> wrong-site surgery. The<br />
complete PIAA database documents almost 1000 closed malpractice claims involving wrong-site<br />
surgery. 6 However, this figure also underestimates the prevalence <strong>of</strong> wrong-site surgery, as every<br />
case does not result in a claim. Most wrong-site surgeries involve relatively minor procedures<br />
such as arthroscopy, 10,15 rather than limb amputations or major neurosurgical procedures.<br />
Consequently sequelae are minimal. The State Volunteer Mutual Insurance Company<br />
(Tennessee) released a series <strong>of</strong> 37 wrong-site surgery claims from 1977 to 1997. 15 Performing<br />
the correct procedure on the wrong side constituted the most common error (eg, arthroscopic<br />
knee surgery on the wrong knee in 15 <strong>of</strong> the 37 cases). Twenty-six <strong>of</strong> the patients experienced no<br />
sequelae beyond a scar, <strong>and</strong> only three patients suffered permanent disability. Given the rarity <strong>of</strong><br />
significant harm, estimates <strong>of</strong> the incidence <strong>of</strong> wrong-site surgery derived from litigation data<br />
likely underestimate the true prevalence <strong>of</strong> this problem, as do estimates based on incident<br />
reports.<br />
Opportunities for Impact<br />
Some surgeons have developed their own methods for surgical site identification, 11,12 but<br />
routine preoperative evaluation <strong>and</strong> marking <strong>of</strong> the intended surgical site by the attending<br />
surgeon has yet to become st<strong>and</strong>ard practice in orthopedics or any surgical specialty. One year<br />
after its "Sign Your Site" campaign, the American Academy <strong>of</strong> Orthopaedic Surgeons surveyed<br />
its membership. Among the 2000 orthopedic surgeons surveyed, 77% responded that the idea<br />
was a good one, but only 40% stated that they complied with the recommended practice. Only<br />
one-third stated that their principal hospitals had a “Sign Your Site” or similar program in place,<br />
although many anticipated initiation <strong>of</strong> one in response to the campaign. 16,17<br />
Study Designs<br />
The published literature includes no studies in which the adoption <strong>of</strong> a practice related to<br />
surgical site identification is analyzed in the setting <strong>of</strong> a controlled observational design or<br />
clinical trial. One report described the experience <strong>of</strong> 4 orthopedic surgeons in private practice, 18<br />
but included no comparable observations from a control group. The experience <strong>of</strong> the Canadian<br />
Orthopaedic Association remains unpublished, <strong>and</strong> the observed effect is based entirely on<br />
litigation statistics 10 (B. Lewis, personal communication, March 2001).<br />
Study Outcomes<br />
Given the egregious <strong>and</strong> distressing nature <strong>of</strong> wrong-site surgery, the error itself<br />
represents the outcome <strong>of</strong> interest, regardless <strong>of</strong> the clinical outcome. Unfortunately, in the<br />
absence <strong>of</strong> observational studies that include controls, the number <strong>of</strong> malpractice claims for<br />
wrong-site surgery represents the most widely cited outcome.<br />
† M<strong>and</strong>atory New York reporting system (NYPORTS) documented 46 events in 24 months.<br />
New York State has a population roughly 1/15 th <strong>of</strong> the entire country. Thus, had JCAHO<br />
captured wrong-site surgeries with the same sensitivity, one would expect 2484 wrong site<br />
surgeries to have been reported during the 87 months covered by the JCAHO sentinel event<br />
policy. This figure is 21 times the actual figure <strong>of</strong> 114 cases.<br />
496
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The Canadian <strong>Medical</strong> Protective Association reported a baseline level <strong>of</strong> litigation for<br />
wrong-site surgery at 7% <strong>of</strong> all orthopedic surgery settlements before 1994 when the Canadian<br />
Orthopaedic Association instituted their “Operate Through Your Initials” policy. 10 Currently,<br />
there are no known wrong-site surgery claims against orthopedic surgeons in Canada. (B. Lewis,<br />
personal communication, March 2001) Interpreting the difference in the rates <strong>of</strong> litigation <strong>of</strong> a<br />
rare occurrence is difficult, however, especially without an accurate estimate <strong>of</strong> the denominator<br />
(ie, the total number <strong>of</strong> relevant procedures performed during the time periods involved).<br />
Moreover, the degree to which Canadian surgeons complied with the policy is unknown. As<br />
mentioned above, only 40% <strong>of</strong> responding American orthopedists reported adoption <strong>of</strong><br />
preoperative site identification in routine practice. 16<br />
In a North Dakota private practice that used preoperative site identification with indelible<br />
ink, there was one incidence <strong>of</strong> wrong-site surgery (pinning <strong>of</strong> the wrong phalanx in a h<strong>and</strong>) in<br />
15,987 consecutive cases. 18 Even assuming that the sole detected case represents the only wrongsite<br />
surgery in this sample, interpreting this low event rate is impossible without a control group.<br />
Comparing this result with national data is also problematic. National data on the number <strong>of</strong><br />
orthopedics procedures performed each year might generate an estimate <strong>of</strong> an appropriate<br />
“denominator,” but the nationwide “numerator” is unknown. Because we do not know the extent<br />
to which incident reporting <strong>and</strong> malpractice litigation underestimate the incidence <strong>of</strong> wrong-site<br />
surgery, we cannot accurately estimate the baseline rate <strong>of</strong> wrong-site surgeries for comparison<br />
with results <strong>of</strong> a case series such as the North Dakota report cited above. 18<br />
Potential for Harm<br />
Some surgeons may worry that marking the surgical site increases the risk <strong>of</strong><br />
contamination, but this concern appears unwarranted. 15,18 More concerning is the potential harm<br />
that may arise from confusion caused by practice variability in “signing the site.” Although the<br />
original recommendations called for surgeons to initial the intended operative site, some<br />
surgeons <strong>and</strong> hospitals mark the site with an “X.” 15,16 Still others use an “X” or “No” to mark the<br />
limb or site that should not be operated upon. 11,12 In addition, there are reports <strong>of</strong> patients<br />
crossing their legs before the ink is dry <strong>and</strong> producing an identical mark on the contralateral<br />
knee, thus subverting the intended effect <strong>of</strong> the intervention. 10 Confusion may also ensue if<br />
operating room personnel cover the mark with surgical drapes prior to the start <strong>of</strong> the surgery. 10<br />
Costs <strong>and</strong> Implementation<br />
The costs <strong>of</strong> marking a surgical site are negligible. Marking procedures that require the<br />
presence <strong>of</strong> the surgeon in the preoperative area prior to the initiation <strong>of</strong> anesthesia may require a<br />
culture shift among surgeons <strong>and</strong> involve the costs <strong>of</strong> surgeons’ time. Implementation strategies<br />
designed to more efficiently utilize the operating surgeon’s time could be designed. For<br />
hospitalized patients, implementation might involve the operating surgeon initialing the intended<br />
operative site at the time consent is obtained, thus requiring that the physician be present at the<br />
time <strong>of</strong> consent. The nurse/anesthetist, anesthesiologist or other responsible party in the preop<br />
area would then be required to contact the operating surgeon only in cases where the operative<br />
site has not already been initialed.<br />
497
Comment<br />
While “signing the site” represents a low-tech solution with high face validity, no<br />
evidence supports a particular version <strong>of</strong> this practice. Additionally, the existence <strong>of</strong> different<br />
versions <strong>of</strong> the “signing the site” practice may cause confusion to the point <strong>of</strong> increasing the<br />
likelihood <strong>of</strong> error. Strategies that focus only on a single aspect <strong>of</strong> the identification problem,<br />
without considering the preoperative <strong>and</strong> operative processes as a whole may fail to avert error.<br />
For instance, protocols that rely on review <strong>of</strong> key preoperative x-rays in the operating room<br />
creates a new m<strong>and</strong>ate to ensure that the correct patient’s x-rays are brought to the operating<br />
room. 6<br />
<strong>Practices</strong> to successfully reduce (<strong>and</strong> eventually eliminate) wrong-site surgeries will<br />
likely combine a st<strong>and</strong>ard method <strong>of</strong> marking the intended site with collaborative protocols for<br />
verification <strong>of</strong> the intended procedure <strong>and</strong> operative site by all members <strong>of</strong> the operating room<br />
staff. 7 Straightforward as each <strong>of</strong> these processes may appear, successful implementation may<br />
require substantial investments <strong>of</strong> time <strong>and</strong> resources for protocol development <strong>and</strong> team<br />
training. Whatever multifaceted practice is developed should be implemented within the setting<br />
<strong>of</strong> a planned observational study.<br />
References<br />
1. <strong>Patient</strong>'s son says family was seeking the best care (family <strong>of</strong> Rajeswari Ayyappan, who<br />
had the wrong side <strong>of</strong> his brain operated on at Memorial Sloan-Kettering Cancer Center,<br />
discusses his health). New York Times. Nov 16 1995: A16(N), B4(L).<br />
2. Olson W. A story that doesn't have a leg to st<strong>and</strong> on (amputation <strong>of</strong> man's left leg rather<br />
than right one may have been medically justified). Wall Street Journal. March 27 1995:<br />
A18(W), A20(E).<br />
3. Altman LK. State fines Sloan-Kettering for an error in brain surgery (New York State fines<br />
Memorial Sloan-Kettering Cancer Center; surgeon operates on wrong side <strong>of</strong> patients<br />
brain). New York Times. March 9 1996: 27(L).<br />
4. Steinhauer J. So, the brain tumor's on the left, right? (Seeking ways to reduce mix-ups in<br />
the operating room; better communication is one remedy, medical experts say). New York<br />
Times. April 1 2001: 23(N), 27(L).<br />
5. American Academy <strong>of</strong> Orthopaedic Surgeons. Advisory statement on wrong-site surgery.<br />
Available at: http://www.aaos.org/wordhtml/papers/advistmt/wrong.htm. Accessed April<br />
16, 2001.<br />
6. Prager LO. Sign here. Surgeons put their signatures on patients' operative sites in an effort<br />
to eliminate the change <strong>of</strong> wrong-site surgeries. Am Med News. October 12 1998: 13-14.<br />
7. The Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Alert.<br />
Lessons Learned: Wrong Site Surgery. Available at: http://www.jcaho.org/edu_pub/<br />
sealert/sea6.html. Accessed March 30, 2001.<br />
8. Lessons learned: sentinel event trends in wrong-site surgery. Jt Comm Perspect.<br />
2000;20:14.<br />
9. Sentinel events: approaches to error reduction <strong>and</strong> prevention. Jt Comm J Qual Improv.<br />
1998;24:175-86.<br />
10. Position paper on wrong-sided surgery in orthopaedics. Winnepeg, Manitoba: Canadian<br />
Orthopaedic Association Committee on Practice <strong>and</strong> Economics; 1994.<br />
11. Lubicky JP. Wrong-site surgery. J Bone Joint Surg Am. 1998;80:1398.<br />
498
12. Cowell HR. Wrong-site surgery. J Bone Joint Surg Am. 1998;80:463.<br />
13. Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations. Sentinel Event Statistics.<br />
Available at: http://www.jcaho.org. Accessed April 16, 2001.<br />
14. U.S. Census Bureau. United States Census 2000: Ranking Tables for States: Population in<br />
2000 <strong>and</strong> Population Change from 1990 to 2000. Available at:<br />
http://www.census.gov/population/www/cen2000/phc-t2.html. Accessed May 31, 2001.<br />
15. American Academy <strong>of</strong> Orthopaedic Surgeons. Report <strong>of</strong> the task force on wrong-site<br />
surgery. Available at: http://www.aaos.org/wordhtml/meded/tasksite/htm. Accessed March<br />
15, 2001.<br />
16. Risk Management Foundation <strong>of</strong> the Harvard <strong>Medical</strong> Institutions. Did Wrong-Site<br />
Surgery Remedy Work? Available at: http://www.rmf.harvard.edu/publications/resource/<br />
feb1999news/article2/. Accessed April 16, 2001.<br />
17. The American Academy <strong>of</strong> Orthopedic Surgeons. Academy News. "Sign your site" gets<br />
strong member support. Available at: http://www.aaos.org/wordhtml/99news/os3-sign.htm.<br />
Accessed May 5, 2001.<br />
18. Johnson PQ. Wrong site surgery in orthopaedics: analysis <strong>of</strong> 15,987 cases at The<br />
Orthopaedic Instititute in Fargo. Paper presented at: American Academy <strong>of</strong> Orthopaedic<br />
Surgeons 67th Annual Meeting; March 15, 2000, 2000; Los Angeles, CA.<br />
499
500
Chapter 44. Crew Resource Management <strong>and</strong> its Applications in Medicine<br />
Laura Pizzi, PharmD<br />
Neil I. Goldfarb<br />
David B. Nash, MD, MBA<br />
Thomas Jefferson University School <strong>of</strong> Medicine<br />
<strong>and</strong> Office <strong>of</strong> Health Policy & Clinical Outcomes<br />
Background<br />
<strong>Patient</strong> care, like other technically complex <strong>and</strong> high risk fields, is an interdependent<br />
process carried out by teams <strong>of</strong> individuals with advanced technical training who have varying<br />
roles <strong>and</strong> decision-making responsibilities. While technical training assures pr<strong>of</strong>iciency at<br />
specific tasks, it does not address the potential for errors created by communicating <strong>and</strong> decision<br />
making in dynamic environments. Experts in aviation have developed safety training focused on<br />
effective team management, known as Crew Resource Management (CRM). Improvements in<br />
the safety record <strong>of</strong> commercial aviation may be due, in part, to this training. 1 Over the past 10<br />
years, lessons from aviation’s approach to team training have been applied to patient safety,<br />
notably in intensive care unit (ICU) <strong>and</strong> anesthesia training. 2,3 This chapter reviews the literature<br />
on Crew Resource Management, also known as Cockpit Resource Management, <strong>and</strong> describes<br />
adaptations <strong>of</strong> this training framework to medicine.<br />
Practice Description<br />
Crew Resource Management in Aviation<br />
Crew Resource Management has been widely used to improve the operation <strong>of</strong> flight<br />
crews. The concept originated in 1979, in response to a NASA workshop that examined the role<br />
that human error plays in air crashes. 4 CRM emphasizes the role <strong>of</strong> human factors in high-stress,<br />
high-risk environments. John K. Lauber, a psychologist member <strong>of</strong> the National Transportation<br />
<strong>Safety</strong> Board, defined CRM as “using all available sources—information, equipment, <strong>and</strong><br />
people—to achieve safe <strong>and</strong> efficient flight operations.” 5,6 CRM encompasses team training, as<br />
well as simulation (also referred to as Line-Oriented Flight Training, or LOFT), interactive<br />
group debriefings, <strong>and</strong> measurement <strong>and</strong> improvement <strong>of</strong> aircrew performance.<br />
There is no universal CRM training program. The Federal Aviation Administration<br />
(FAA) allows air carriers to customize their CRM programs to best suit the needs <strong>of</strong> individual<br />
organizations. Therefore, training programs vary somewhat from carrier to carrier, making it<br />
difficult to describe operational components. Furthermore, these programs continue to evolve as<br />
aviation technology changes <strong>and</strong> more is learned about group dynamics.<br />
One CRM model focuses on the elements <strong>of</strong> human effectiveness. 7 The 3 primary<br />
components <strong>of</strong> effective crew management are safety, efficiency, <strong>and</strong> morale. Specific factors<br />
related to aircrew performance are categorized, <strong>and</strong> serve as the basis for training <strong>and</strong> research.<br />
These factors include materials, organization, individual, <strong>and</strong> group process variables associated<br />
with performance. 8 Examples <strong>of</strong> outcomes that result from these input variables are safety,<br />
efficiency, <strong>and</strong> customer satisfaction.<br />
Subsequently, Helmreich <strong>and</strong> colleagues proposed a modified conceptual framework for<br />
CRM, termed the “Error Troika,” 9 to display a hierarchy <strong>of</strong> 3 error countermeasures. At the first<br />
501
level, CRM includes training on how to avoid errors. At the second level, potential errors are<br />
“trapped” before they are committed. At the third level, mitigation <strong>of</strong> error consequences occurs.<br />
From a practical st<strong>and</strong>point, CRM programs typically include educating crews about the<br />
limitations <strong>of</strong> human performance. 10 Trainees develop an underst<strong>and</strong>ing <strong>of</strong> cognitive errors, <strong>and</strong><br />
how stressors (such as fatigue, emergencies, <strong>and</strong> work overload) contribute to the occurrence <strong>of</strong><br />
errors. Multi-day CRM training programs typically require participants to assess personal <strong>and</strong><br />
peer behavior. Operational concepts stressed include inquiry, seeking relevant operational<br />
information, advocacy, communicating proposed actions, conflict resolution <strong>and</strong> decision<br />
making.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
The field <strong>of</strong> aviation has a substantial history <strong>of</strong> collecting <strong>and</strong> analyzing safety-related<br />
data. Historically, human error has caused or contributed to over 50% <strong>of</strong> aviation accidents. In an<br />
analysis <strong>of</strong> 35,000 reports <strong>of</strong> incidents over 7.5 years, almost 50% resulted from a flight crew<br />
error, <strong>and</strong> an additional 35% were attributed to air traffic controller error. 11 Root cause analyses<br />
(Chapter 5) by safety experts have found that errors frequently occur because flight crews fail to<br />
effectively manage the resources available to them (eg, fail to verify information when uncertain<br />
about it, fail to plan for contingencies). 11 Naval aviation reports provide similar results, with one<br />
study reporting 59% <strong>of</strong> “Class A mishaps” (serious consequences including fatality, destroyed<br />
aircraft, <strong>and</strong> major injury) attributed to some degree to aircrew factors. 12 These <strong>and</strong> similar<br />
analyses have catalyzed tailored prevention strategies including CRM for commercial aviation, 1<br />
<strong>and</strong> “aircrew coordination training” (ACT) for Naval aviators. 12<br />
Study Design <strong>and</strong> Outcomes<br />
Measures<br />
Although the most obvious <strong>and</strong> meaningful measure <strong>of</strong> CRM effectiveness would appear<br />
to be airline accident or “near miss” rates, these objective measures have not been used in<br />
commercial aviation studies. Helmreich suggests that it is not possible to use these measures,<br />
because accident rates are very low <strong>and</strong> “near misses” are voluntarily reported. 10 Furthermore,<br />
the content <strong>and</strong> structure <strong>of</strong> CRM training programs are variable. 10<br />
In response, researchers have developed tools that assess the effectiveness <strong>of</strong> CRM in<br />
other ways. These tools include attitudinal surveys <strong>and</strong> peer performance rating questionnaires,<br />
including the NASA/University <strong>of</strong> Texas Line/LOS Checklist (LINE/LOS Checklist), 13 the<br />
Cockpit Management Attitudes Questionnaire (CMAQ), <strong>and</strong> the Flight Management Attitudes<br />
Questionnaire (FMAQ). The LINE/LOS Checklist is used to rate crew performance on critical<br />
behaviors during specific segments <strong>of</strong> flight (eg, not rushing through briefing period, exhibiting<br />
high levels <strong>of</strong> vigilance in both high <strong>and</strong> low workload conditions, etc). Ratings on each<br />
behavioral element (ie, model for teamwork) range across 4 levels from poor to outst<strong>and</strong>ing.<br />
In contrast, CMAQ is used to evaluate the attitudes <strong>of</strong> crewmembers within <strong>and</strong> between<br />
organizations, pre- <strong>and</strong>/or post-CRM training. Results are intended to serve as a proxy for<br />
measuring crew process <strong>and</strong> performance. 14 The instrument has been validated by comparing<br />
self-reported attitudes with performance ratings made by experienced Check Airmen, experts<br />
trained in peer evaluation. 15 The FMAQ is a revised version <strong>of</strong> the CMAQ that was developed by<br />
Helmreich <strong>and</strong> colleagues in response to attitudinal differences observed in flight crews from<br />
different countries. 16<br />
Observation <strong>of</strong> crew performance in simulated flights has also been used.<br />
502
Representative Studies<br />
Several studies have utilized proxy tools to test the effectiveness <strong>of</strong> CRM. 8,12,17 One study<br />
by Helmreich <strong>and</strong> colleagues consisted <strong>of</strong> an assessment <strong>of</strong> the attitudes before versus after CRM<br />
training (pre-test versus post-test). 17 Crew behaviors were noted by a trained observer using the<br />
NASA/University <strong>of</strong> Texas Line/LOS Checklist. 18 More than 2000 line flights <strong>and</strong> LOFT<br />
sessions were included in the analysis. Overall performance <strong>of</strong> crews was classified as “below<br />
average,” “average,” or “above average” by Check Airmen <strong>and</strong> LOFT instructors.<br />
As a result <strong>of</strong> the CRM training, the percentage <strong>of</strong> crews rated as “above average”<br />
increased while the percent rated “below average” decreased. Performance ratings differed<br />
between fleets <strong>and</strong> airlines. Superior pilots also shared many common attitudes, (for example,<br />
they were aware <strong>of</strong> their personal limitations <strong>and</strong> diminished decision-making capacity during<br />
emergencies). In addition, they encouraged crewmembers to question their decisions <strong>and</strong> actions,<br />
were sensitive to the personal problems <strong>of</strong> other crewmembers, <strong>and</strong> recognized the need to<br />
verbalize plans <strong>and</strong> to train other crewmembers.<br />
A second study by Barker <strong>and</strong> colleagues compared the effectiveness <strong>of</strong> 17 CRM-trained<br />
flight crews on military mission simulators, divided according to whether they were “fixed” or<br />
“formed” crews. 8 Nine <strong>of</strong> the crews were defined as "fixed" since they had flown together for six<br />
months or longer; the remaining 8 were “formed,” as they had flown together for less than 6<br />
months. Each crew was asked to participate in a simulated mission. The first leg <strong>of</strong> the mission<br />
was programmed to be problem-free, but the second leg required crews to address a safety issue.<br />
As with the earlier study, crew behaviors were observed using the NASA/University <strong>of</strong> Texas<br />
Line/LOS Checklist. Surprisingly, the formed crews committed fewer minor errors, but the<br />
number <strong>of</strong> major errors did not differ significantly between the groups. 8 The authors concluded<br />
that formed crews may experience less ineffective coordination than fixed crews, as the latter<br />
might be more complacent from the routine <strong>of</strong> working together, but that further research was<br />
needed.<br />
A third study evaluated rates <strong>of</strong> naval aviation mishaps, <strong>and</strong> the role <strong>of</strong> CRM failures. 12<br />
While the study primarily compared crew performance in 2 types <strong>of</strong> equipment, the aircrew<br />
performance deficits were also compared with those seen during an earlier time period, prior to<br />
implementation <strong>of</strong> aircrew coordinating training (ACT) programs (the specific form <strong>of</strong> CRM<br />
used). <strong>Analysis</strong> <strong>of</strong> data from the post-ACT implementation period across the fleet revealed that<br />
70% <strong>of</strong> human error mishaps were connected with aircrew factors, <strong>and</strong> that 56% <strong>of</strong> these resulted<br />
from at least one CRM failure. The authors noted that these percentages were similar to those<br />
reported in a separate study prior to ACT implementation. Because <strong>of</strong> a lack <strong>of</strong> controls or<br />
adjustments for potential confounders, no conclusions about effectiveness <strong>of</strong> the ACT program<br />
can be drawn. 12<br />
503
Comparison to Medicine<br />
Sexton <strong>and</strong> colleagues compared flight crews with operating room personnel on several<br />
measures, including attitudes toward teamwork. 19 The study included more than 30,000 cockpit<br />
crew members (captains, first <strong>of</strong>ficers, <strong>and</strong> second <strong>of</strong>ficers) <strong>and</strong> 1033 operating room personnel<br />
(attending surgeons, attending anesthesiologists, surgical residents, anesthesia residents, surgical<br />
nurses, anesthesia nurses). Data from the crew members were obtained from previous<br />
administrations <strong>of</strong> the CMAQ <strong>and</strong> FMAQ to major airlines around the world (over a 15-year<br />
period). The operating room participants were mailed an analogous questionnaire (CMAQ<br />
modified 20 ), administered over a period <strong>of</strong> 3 years at 12 teaching <strong>and</strong> non-teaching hospitals in<br />
the United States <strong>and</strong> abroad (Italy, Germany, Switzerl<strong>and</strong>, <strong>and</strong> Israel).<br />
The level <strong>of</strong> teamwork perceived by attending surgeons compared with other operating<br />
room staff differed markedly. A majority <strong>of</strong> surgical residents (73%) <strong>and</strong> attending surgeons<br />
(64%) reported high levels <strong>of</strong> teamwork, but only 39% <strong>of</strong> attending anesthesiologists, 28% <strong>of</strong><br />
surgical nurses, 25% <strong>of</strong> anesthesia nurses, <strong>and</strong> 10% <strong>of</strong> anesthesia residents reported high levels<br />
<strong>of</strong> teamwork. A bare majority (55%) <strong>of</strong> attending surgeons rejected steep hierarchies (determined<br />
by whether they thought junior team members should question the decisions <strong>of</strong> senior team<br />
members). In contrast, 94% <strong>of</strong> airline crew members preferred flat hierarchies.<br />
It was also noted that medical participants were far more likely to agree with the<br />
statement “Even when fatigued, I perform effectively during critical times.” Seventy percent <strong>of</strong><br />
attending surgeons agreed with this statement, as well as 56% <strong>of</strong> surgical residents, 60% <strong>of</strong><br />
surgical nurses, 57% <strong>of</strong> anesthesia residents, 55% <strong>of</strong> anesthesia nurses, <strong>and</strong> 47% <strong>of</strong> attending<br />
anesthesiologists. In contrast, 26% <strong>of</strong> pilots agreed with this statement.<br />
Applications <strong>of</strong> CRM Principles in Medicine<br />
The Sexton study <strong>and</strong> other analyses suggest that safety-related behaviors that have been<br />
applied <strong>and</strong> studied extensively in the aviation industry may also be relevant in health care. We<br />
identified CRM applications in several dynamic decision-making health care environments: the<br />
operating room, labor <strong>and</strong> delivery, <strong>and</strong> the emergency room. 3,21,22 In addition, Gaba has noted<br />
that some other domains (eg, cardiac arrest response teams) that have active simulation training<br />
are currently incorporating a broader range <strong>of</strong> CRM-like training methods. 23 (Simulators are<br />
covered in more detail in Chapter 45).<br />
Practice Description<br />
Crew Resource Management in Health Care Settings<br />
As with aviation, the medical application <strong>of</strong> CRM has required tailoring <strong>of</strong> training<br />
approaches to mirror the areas in which human factors contribute to mishaps. In anesthesiology<br />
65-70% <strong>of</strong> safety problems (accidents or incidents) have been attributed at least in part to human<br />
error. In response, several anesthesiologists from the VA Palo Alto Health Care System <strong>and</strong><br />
Stanford University, with funding from the Anesthesia <strong>Patient</strong> <strong>Safety</strong> Foundation, developed<br />
Anesthesia Crisis Resource Management (ACRM), modeled on CRM. 3,23 The original<br />
demonstration courses consisted <strong>of</strong> didactic instruction, videotape <strong>of</strong> a reenactment <strong>of</strong> an<br />
aviation disaster, videotape <strong>of</strong> an actual anesthetic mishap, simulation training <strong>and</strong> a debriefing<br />
session. Ongoing courses include the use <strong>of</strong> a textbook that catalogues 83 critical events (eg,<br />
acute hemorrhage, bronchospasm, seizures), <strong>and</strong> approaches to managing them. 24 Currently,<br />
there are 3 ACRM courses <strong>of</strong>fering progressively more challenging material, a Working Group<br />
504
on Crisis Management Training in Health Care formed by the developers <strong>of</strong> ACRM that has<br />
initiated formal ACRM instructor training, <strong>and</strong> thoughtful consideration <strong>of</strong> pragmatic approaches<br />
to evaluating ACRM. 23<br />
Helmreich <strong>and</strong> Schaefer have also advanced CRM theory in the operating room<br />
environment by adapting their model <strong>of</strong> team performance. This framework describes the team<br />
performance inputs that are critical to essential team functions that in turn lead to desired<br />
outcomes (defined as patient well-being). Examples <strong>of</strong> inputs include individual aptitudes,<br />
physical environment, <strong>and</strong> culture (pr<strong>of</strong>essional, organizational, <strong>and</strong> national). Performance<br />
functions consist <strong>of</strong> team formation <strong>and</strong> management, surgical procedures, communications,<br />
decision processes, <strong>and</strong> situational awareness. 18<br />
Another application <strong>of</strong> CRM to the health environment is the MedTeams behavior-based<br />
teamwork system, developed by Dynamics Research Corporation, <strong>and</strong> sponsored by the Army<br />
Research Laboratory. It aims to adapt research in team performance <strong>and</strong> training from military<br />
helicopter aviation to emergency medicine. 22,25,26 Thus far, specific applications have been<br />
developed for Emergency Department care <strong>and</strong> labor <strong>and</strong> delivery units. 27 The system is<br />
implemented through courses <strong>and</strong> assessment tools. The Emergency Team Coordination Course<br />
(ETCC) includes 5 team dimensions or goals (ie, maintain team structure <strong>and</strong> climate, facilitate<br />
planning <strong>and</strong> problem-solving, enhance communication among team members, facilitate<br />
workload management, improve team-building skills). Each goal is tied to specific teamwork<br />
tasks. For example, tasks for the first goal (maintain team structure <strong>and</strong> climate) include<br />
“establish team leader,” “form the team,” “set team goals,” <strong>and</strong> “assign roles <strong>and</strong><br />
responsibilities.” 22 Like the CRM approach to the “Error Troika,” the MedTeams approach is<br />
based on avoiding errors, trapping them as they occur, <strong>and</strong> mitigating the consequences <strong>of</strong> actual<br />
errors. Principles underlying the MedTeams approach include:<br />
• Team responsibility for patients<br />
• A belief in clinician fallibility<br />
• Peer monitoring<br />
• Team member awareness <strong>of</strong> patient status, team member status <strong>and</strong><br />
institutional resources<br />
Peer monitoring is a fundamental component <strong>of</strong> the MedTeams system, as well as a<br />
feature <strong>of</strong> the ACRM <strong>and</strong> aviation field’s CRM approaches. Along with his or her clinical<br />
responsibilities, each team member undertakes the intermittent process <strong>of</strong> peer monitoring or<br />
"check" actions, engaging in this check cycle as frequently as possible. The teamwork check<br />
cycle begins with each team member monitoring his or her own situation awareness <strong>and</strong> crossmonitoring<br />
the actions <strong>of</strong> other teammates. If during the monitoring mode the monitoring<br />
teammate observes a suspected error in progress, that individual intervenes with a direct question<br />
or <strong>of</strong>fer <strong>of</strong> information. The erring teammate may then acknowledge the lapse, correct it <strong>and</strong><br />
continue working. Alternatively, the monitoring teammate may have lost situation awareness.<br />
The non-erring monitored colleague can then provide feedback to correct the peer's situation<br />
awareness. If team members are in strong disagreement about how patient care should proceed,<br />
advocacy, assertion <strong>and</strong> perhaps third-party involvement may be used to resolve the situation.<br />
Over time, the check cycle becomes habitual, resulting in hundreds <strong>of</strong> team checks daily, all with<br />
the potential to break the error chain.<br />
Recently, ACRM has been extended to a full-day course on neonatal resuscitation<br />
training for neonatologists <strong>and</strong> pediatricians. 21 Called “NeoSim”, the course combines traditional<br />
505
training methods with reviews <strong>of</strong> literature <strong>and</strong> didactic instruction with simulation. Debriefing<br />
follows, using videotape <strong>of</strong> the simulation. As with the other examples, the emphasis <strong>of</strong> teaching<br />
behavioral teamwork skills along with technical content is the hallmark <strong>of</strong> CRM interventions in<br />
health care.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The most thoroughly studied <strong>of</strong> the medical CRM applications is ACRM, although as<br />
with the aviation examples, rigorous evaluations are challenging to design. Few studies utilize a<br />
control group, although researchers have reported assessment methods for determining both<br />
technical <strong>and</strong> behavioral performance from simulator videotapes 28 (see also Chapter 45). A<br />
before-after ACRM training analysis (Level 3 Study Design) <strong>of</strong> trainees’ knowledge-base (Level<br />
3 Outcome) for crisis yielded mixed results. 3 The average score on the post-test was significantly<br />
greater than the pre-test for one trainee class, composed mostly <strong>of</strong> residents. For the other class<br />
<strong>of</strong> experienced anesthesiologists, the test scores did not change <strong>and</strong> were at the same level as the<br />
post-test scores <strong>of</strong> class <strong>of</strong> residents. Subjective data evaluating the course indicated that trainees<br />
“uniformly felt that the ACRM course was an intense, superior form <strong>of</strong> training related to an<br />
important, but inadequately taught, component <strong>of</strong> anesthesia practice.” 3 Another study <strong>of</strong> ACRM<br />
at Harvard also found that participants rated the course favorably, with over 80% responding that<br />
they felt the course should be taken every 24 months or less. 29<br />
As with aviation, the incremental value <strong>of</strong> this form <strong>of</strong> training is difficult to link to<br />
improvements in teamwork performance <strong>and</strong> better safety records. At the time <strong>of</strong> this literature<br />
assessment, there were no published data to describe the effects on medical error rates <strong>of</strong> the<br />
MedTeams approach. The NeoSim course participants provided positive responses to openended<br />
questions about their satisfaction with the course. 21<br />
Costs <strong>and</strong> Implementation<br />
Helmreich has noted some <strong>of</strong> the limitations associated with CRM. 10 At this time, the<br />
evidence connecting CRM approaches to improving patient safety does not exist,<br />
notwithst<strong>and</strong>ing the face validity <strong>of</strong> the approach. Nevertheless, a long history <strong>of</strong> variants <strong>of</strong> the<br />
approach <strong>of</strong>fers health care a reasonable foundation from which to draw practical <strong>and</strong> evidencedbased<br />
resources 30 for further development <strong>and</strong> adaptation <strong>of</strong> CRM, as well as measurement<br />
methods to ascertain its effectiveness.<br />
At a minimum, implementation <strong>of</strong> the CRM approach in health care settings requires<br />
customization <strong>of</strong> tools <strong>and</strong> techniques for each specific care venue, as is illustrated by<br />
adaptations implemented thus far. This customization comes at considerable cost <strong>and</strong> cannot be<br />
expected to immediately reap safety benefits. At the time <strong>of</strong> this review, approximate costs for<br />
implementing the MedTeams system ranged from $15,000-$35,000, with additional costs for<br />
ongoing activities (such as continuing education) that ranged from $8,000-$20,000 (R. Simon,<br />
personal communication, April 2001). Similarly, marginal costs for CRM-like training based on<br />
the ACRM experience are estimated at $800 to $2,000 per participant per day (D. M. Gaba,<br />
personal communication, June 2001). These costs do not include the overhead <strong>of</strong> starting a<br />
program (eg, simulator investment, training instructors), nor do they factor in the cost <strong>of</strong> reduced<br />
practice work hours, if these are above those devoted to current training time.<br />
Cultural shifts in medicine are also necessary if the CRM approach is truly to take root.<br />
CRM applications are relatively novel in health care, a field in which pr<strong>of</strong>essional training <strong>and</strong><br />
education have traditionally focused on developing technical pr<strong>of</strong>iciency rather than facilitating<br />
human interaction. Although communication <strong>and</strong> decision making are central to medical<br />
506
practice, relatively little about this topic has appeared in medical literature, despite the fact that<br />
information flow is critical, particularly in high acuity venues such as the operating room <strong>and</strong><br />
emergency departments. Ideas must be elicited, debated <strong>and</strong> evaluated without discrimination<br />
based on the status <strong>of</strong> the staff person <strong>of</strong>fering the information. 31<br />
Paradoxically, the attachment to hierarchy may be the reason that a small percentage <strong>of</strong><br />
participants can be expected to reject CRM training. Research has shown that some resistance is<br />
rooted in personality characteristics. Crew members lacking in achievement motivation <strong>and</strong><br />
interpersonal skills are more likely to reject the training. Additionally, CRM practices decay over<br />
time, even with repeated training, requiring continuing expenditures to retain the hoped-for<br />
gains. 10<br />
Comment<br />
CRM has evolved in the airline industry for more than 20 years, <strong>and</strong> has been extensively<br />
applied during the past decade. Although no definitive data link CRM to decreased aviation error<br />
per se, the industry has accepted the face validity <strong>of</strong> the practice, <strong>and</strong> it is now an integral part <strong>of</strong><br />
training. Over time, survey data from thous<strong>and</strong>s <strong>of</strong> civilian <strong>and</strong> military participants in the<br />
United States <strong>and</strong> abroad has been accrued. These data indicate that most flight crew members<br />
accept CRM training, <strong>and</strong> find it both relevant <strong>and</strong> useful. 7<br />
The studies reviewed provide some support for the notion that CRM is worth further<br />
investigation in health care. However, it cannot yet be concluded that CRM is a practice that can<br />
reduce medical errors. Additional research in this area is warranted, although measurement <strong>and</strong><br />
study design are particularly challenging. As noted, although the evidence from aviation, where<br />
CRM is well-established, has shortcomings, data are easier to capture since CRM assessments<br />
can be grafted onto m<strong>and</strong>atory, ongoing, yearly pilot assessments conducted both in simulators<br />
<strong>and</strong> in real flights (D. M. Gaba, personal communication, June 2001). In medicine, where<br />
ongoing assessments are not the norm, capturing relevant data will be more difficult logistically<br />
<strong>and</strong> much more expensive than in aviation. Consequently, <strong>and</strong> for the analogous reasons that<br />
aviation has adopted CRM based on face validity, health care decision makers may wish to<br />
consider face validity in lieu <strong>of</strong> massive research investments.<br />
Nonetheless, evaluations <strong>of</strong> CRM that focus on intermediate outcomes (eg, trainee<br />
performance) are feasible, <strong>and</strong> instructive for optimizing components <strong>of</strong> CRM programs. CRM<br />
design <strong>and</strong> evaluation resources are becoming more widely available, 30 <strong>and</strong> health care should<br />
continue to consult these sources, building on the advances made in other fields.<br />
Acknowledgements<br />
The authors are grateful to Robert Simon, Chief Scientist at Dynamics Research<br />
Corporation, for providing details on the MedTeams approach.<br />
References<br />
1. Helmreich RL. On error management: lessons from aviation. BMJ. 2000; 320: 781-5.<br />
2. Shortell SM, Zimmerman JE, Rousseau DM, Gillies RR, Wagner DP, Draper EA, et al.<br />
The performance <strong>of</strong> intensive care units: does good management make a difference? Med<br />
Care. 1994; 32: 508-25.<br />
3. Howard SK, Gaba DM, Fish KJ, Yang G, Sarnquist FH. Anesthesia crisis resource<br />
management training: teaching anesthesiologists to h<strong>and</strong>le critical incidents. Aviat Space<br />
Environ Med. 1992; 63: 763-70.<br />
507
4. Cooper GE, White MD, Lauber JK. Resource management on the flightdeck: proceedings<br />
<strong>of</strong> a NASA/ Industry Workshop. M<strong>of</strong>fett Field, Calif: NASA - Ames Research Center; 1980.<br />
NASA Conference <strong>Public</strong>ation No. CP-2120.<br />
5. Lauber JK. Cockpit resource management: background <strong>and</strong> overview. In: Orlady HW,<br />
Foushee HC, eds. Cockpit resource management training: proceedings <strong>of</strong> the NASA/MAC<br />
workshop. M<strong>of</strong>fett Field, Calif: NASA - Ames Research Center. NASA Conference<br />
<strong>Public</strong>ation No. 2455.<br />
6. Wiener EL, Kanki BG, Helmreich RL. Cockpit resource management. San Diego, Calif:<br />
Academic Press, Inc.; 1993.<br />
7. Helmreich RL, Foushee HC. Why crew resource management? Empirical <strong>and</strong> theoretical<br />
bases <strong>of</strong> human factors in training <strong>and</strong> aviation. In: Wiener E, Kanki BG, Helmreich RL,<br />
eds. Cockpit resource management. San Diego, Calif: Academic Press; 1993: 3-45.<br />
8. Barker JM, Clothier CC, Woody JR, McKinney EH, Jr., Brown JL. Crew resource<br />
management: a simulator study comparing fixed versus formed aircrews. Aviat Space<br />
Environ Med. 1996; 67: 3-7.<br />
9. Helmreich RL, Merritt AC. Cultural issues in crew resource management. Conference<br />
presentation at the ICAO Global Human Factors Seminar; April, 1996; Auckl<strong>and</strong>, New<br />
Zeal<strong>and</strong>.<br />
10. The evolution <strong>of</strong> crew resource management training in commercial aviation. Available at:<br />
http://www.psy.utexas.edu/psy/helmreich/Evolution_IJAP_for_Dist.htm. Accessed June<br />
18, 2001.<br />
11. Billings CE, Reynard WD. Human factors in aircraft incidents: results <strong>of</strong> a 7-year study.<br />
Aviat Space Environ Med. 1984; 55: 960-5.<br />
12. Wiegmann DA, Shappell SA. Human error <strong>and</strong> crew resource management failures in<br />
Naval aviation mishaps: a review <strong>of</strong> U.S. Naval <strong>Safety</strong> Center data, 1990-96. Aviat Space<br />
Environ Med. 1999; 70: 1147-51.<br />
13. Helmreich RL, Wilhelm JA, Kello JE, Taggart WR, Butler RE. Reinforcing <strong>and</strong> evaluating<br />
crew resource management: Evaluator/LOS instructor reference manual. Austin, Tex:<br />
NASA - University <strong>of</strong> Texas at Austin; 1990. Technical Manual 90-2.<br />
14. Gregorich SE, Helmrich RL, Wilhelm JA. The structure <strong>of</strong> cockpit-management attitudes.<br />
J Appl Psychol. 1990; 75: 682-90.<br />
15. Helmreich RL, Foushee BR, Benson R, Russini W. Cockpit Resource Management:<br />
Exploring the attitude-performance linkage. Aviat Space Environ Med. 1986; 57: 1198-200.<br />
16. Helmreich RL, Merritt AC, Sherman PJ, Gregorich SE, Wiener EL. The flight management<br />
attitudes questionnaire. Austin, Tex: NASA/University <strong>of</strong> Texas/FAA; 1993. Technical<br />
Report 93-5.<br />
17. Helmreich RL, Wilhelm JA, Gregorich SE, Chidester TR. Preliminary results from the<br />
evaluation <strong>of</strong> cockpit resource management training: performance ratings <strong>of</strong> flight crews.<br />
Aviat Space Environ Med. 1990; 61: 576-9.<br />
18. Helmreich RL, Schaefer HG. Team performance in the operating room. In: Bogner MS, ed.<br />
Human error in medicine. Hillside, NJ: Lawrence Erlbaum; 1998.<br />
19. Sexton JB, Thomas EJ, Helmreich RL. Error, stress, <strong>and</strong> teamwork in medicine <strong>and</strong><br />
aviation: cross sectional surveys. BMJ. 2000; 320: 745-9.<br />
20. Helmreich RL, Merritt AC. Culture at work in aviation <strong>and</strong> medicine: national,<br />
organizational, <strong>and</strong> pr<strong>of</strong>essional influences. Hants, Engl<strong>and</strong>: Ashgate Publishing Limited;<br />
1998.<br />
508
21. Halamek LP, Kaegi DM, Gaba DM, Sowb YA, Smith BC, Smith BE, et al. Time for a new<br />
paradigm in pediatric medical education: teaching neonatal resuscitation in a simulated<br />
delivery room environment. Pediatrics. 2000; 106: E45.<br />
22. Risser DT, Rice MM, Salisbury ML, Simon R, Jay GD, Berns SD. The potential for<br />
improved teamwork to reduce medical errors in the emergency department. The MedTeams<br />
Research Consortium. Ann Emerg Med. 1999; 34: 373-83.<br />
23. Gaba DM, Howard SK, Fish KJ, Yasser SA. Simulation-based training in Anesthesia Crisis<br />
Resource Management (ACRM): a decade <strong>of</strong> experience. In: Simulation & Gaming. Vol<br />
32: Sage <strong>Public</strong>ations, Inc. In Press.<br />
24. Gaba DM, Fish KJ, Howard SK. Crisis Management in Anesthesiology.: Churchill<br />
Livingstone Inc.; 1994.<br />
25. Small SD, Wuerz RC, Simon R, Shapiro N, Conn A, Setnik G. Demonstration <strong>of</strong> highfidelity<br />
simulation team training for emergency medicine. Acad Emerg Med. 1999; 6: 312-<br />
23.<br />
26. Dynamics Research Corporation. Training analysis <strong>and</strong> delivery. Available at:<br />
http://www.drc.com/Training<strong>Analysis</strong>/medteams.htm. Accessed June, 2001.<br />
27. Spath PL, ed. Error reduction in health care: a systems approach to improving patient<br />
safety. San Francisco, Calif: Jossey-Bass Publishers; 1999.<br />
28. Gaba DM, Howard SK, Flanagan B, Smith BE, Fish KJ, Botney R. Assessment <strong>of</strong> clinical<br />
performance during simulated crises using both technical <strong>and</strong> behavioral ratings.<br />
Anesthesiology. 1998; 89: 8-18.<br />
29. Holzman RS, Cooper JB, Gaba DM, Philip JH, Small SD, Feinstein D. Anesthesia crisis<br />
resource management: real-life simulation training in operating room crises. J Clin Anesth.<br />
1995; 7: 675-87.<br />
30. Salas E, Rhodenizer L, Bowers CA. The design <strong>and</strong> delivery <strong>of</strong> crew resource management<br />
training: exploiting available resources. Hum Factors. 2000; 42: 490-511.<br />
31. Schenkel S. Promoting patient safety <strong>and</strong> preventing medical error in emergency<br />
departments. Acad Emerg Med. 2000; 7: 1204-22.<br />
509
510
Chapter 45. Simulator-Based Training <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong><br />
Ashish K. Jha, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Bradford W. Duncan, MD<br />
Stanford University School <strong>of</strong> Medicine<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Background<br />
For a number <strong>of</strong> years, simulators have been used in aviation, nuclear power, military<br />
flight operations <strong>and</strong> other industries as a training tool <strong>and</strong> method to assess performance. Their<br />
use is nearly universal in high reliability organizations. 1 Recently the use <strong>of</strong> simulation in<br />
medicine has increased markedly, in part due to greater awareness <strong>of</strong> the importance <strong>of</strong> patient<br />
safety.<br />
Defined broadly, a simulator replicates a task environment with sufficient realism to<br />
serve a desired purpose. 1 In medical training, simulators can substitute for actual patients <strong>and</strong> can<br />
be as simple as utilizing pigs’ feet to practice suturing, or as complex as virtual reality machines<br />
<strong>and</strong> re-creations <strong>of</strong> actual clinical environments for surgeons, radiologists <strong>and</strong> anesthesiologists.<br />
In a general sense, they improve patient safety by allowing physicians to become better trained<br />
without putting patients at risk. For example, in a r<strong>and</strong>omized controlled trial, Peugnet <strong>and</strong><br />
colleagues used a virtual reality simulator to train physicians to perform retinal photocoagulation<br />
<strong>and</strong> found that surgeons who trained with the simulator performed the procedure as well as those<br />
who trained with patients. 2 Gaba lists several other advantages <strong>of</strong> simulation, among them 3 :<br />
• Presentation <strong>of</strong> uncommon but critical scenarios in which a rapid response is<br />
needed (eg, malignant hyperthermia, which occurs once in every 40,000<br />
anesthesia cases). 4 To conduct systematic training about managing such<br />
critical events there is little alternative but to use simulation.<br />
• Errors can be allowed to occur <strong>and</strong> reach their conclusion—in real life a more<br />
capable clinician would have to intervene—so participants can see the results<br />
<strong>of</strong> their decisions <strong>and</strong> actions.<br />
• With mannequin-based simulators clinicians can use actual medical<br />
equipment, exposing limitations in the human-machine interface.<br />
• Complete interpersonal interactions with other clinical staff can be explored<br />
<strong>and</strong> training in teamwork, leadership, <strong>and</strong> communication provided. In a<br />
number <strong>of</strong> medical fields, simulation has been used in crew resource<br />
management (CRM) training 1 (see Chapter 44), where the focus is on<br />
behavioral skills such as inter-team communication during critical incidents.<br />
Human error is a leading cause <strong>of</strong> adverse anesthesia events that lead to poor patient<br />
outcomes 5 <strong>and</strong> may play an important role in surgical errors as well. 6 However, it may be<br />
difficult to demonstrate improved patient outcomes from simulation because adverse events are<br />
unusual <strong>and</strong> there are an extreme number <strong>of</strong> potential confounders. Given these difficulties,<br />
performance has been used as a surrogate outcome, despite concerns that it may be a less than<br />
perfect measure. It is yet unclear which attributes <strong>of</strong> performance matter most to patient<br />
outcomes. Additionally, the methods (eg, reliability <strong>and</strong> consistency) <strong>and</strong> timing (eg, duration <strong>of</strong><br />
511
follow-up) by which performance is measured also remain ill-defined. 1 Studies <strong>of</strong> the<br />
effectiveness <strong>of</strong> simulators <strong>of</strong>ten are limited in that they measure performance using the same<br />
training simulator, which may favor those who have trained on the simulator itself. In other<br />
words, seemingly improved performance may not translate to actual patient care. Of the studies<br />
that have extended laboratory simulation to patient care, few have evaluated the impact on<br />
medical error or established a clear link between simulator training <strong>and</strong> patient outcomes.<br />
This chapter reviews the evidence regarding the use <strong>of</strong> simulators in the training <strong>and</strong><br />
ongoing education <strong>of</strong> health care providers as a patient safety measure. Not included in this<br />
review is the use <strong>of</strong> simulators in the planning or preparation <strong>of</strong> diagnostic or therapeutic<br />
interventions in specific patients (eg, calculation <strong>of</strong> radiotherapy doses using computer<br />
simulations, planning a surgical procedure for a specific patient using 3-D simulations based on<br />
the anatomical data from the actual patient). While these kinds <strong>of</strong> simulation clearly decrease the<br />
morbidity <strong>and</strong> mortality associated with their particular procedures, they are omitted here<br />
because they focus on unique characteristics <strong>of</strong> individual patients that may not be generalizable<br />
to other patients. Other uses <strong>of</strong> simulators beyond the scope <strong>of</strong> this discussion include simulators<br />
to measure worker pr<strong>of</strong>iciency, 7-10 to identify areas for educational intervention, <strong>and</strong> to target<br />
improvements in training. 11,12<br />
Simulators in Anesthesia<br />
<strong>Patient</strong> simulators have been most widely studied in anesthesia where human error may<br />
account for over 80% <strong>of</strong> critical incidents. 5 Simulators range from simple mannequins to highfidelity<br />
simulators that recreate the operating room experience. According to one study, 71% <strong>of</strong><br />
medical schools in Canada, the United Kingdom <strong>and</strong> other western nations used mannequins or<br />
some other form <strong>of</strong> simulator to teach anesthesia to medical students. 13 There is a growing body<br />
<strong>of</strong> literature on the different roles that simulators can play in anesthesia training. 14<br />
Studies have found that simulators can effectively identify errors <strong>and</strong> appropriateness <strong>of</strong><br />
decision making in anesthesia. For example, during 19 comprehensive anesthesia simulations,<br />
DeAnda et al documented 132 unplanned incidents, <strong>of</strong> which 87 (66%) were due to human error<br />
<strong>and</strong> 32 (27%) were considered critical incidents. 15 Schwid <strong>and</strong> O’Donnell have also used<br />
simulators to document the type <strong>of</strong> errors that anesthesiologists make in critical incidents,<br />
finding errors in monitor usage (37%), airway management (17%), ventilator management<br />
(13%), <strong>and</strong> drug administration (10%). 16 Gaba <strong>and</strong> colleagues studied both the appropriateness<br />
<strong>of</strong> decisions <strong>and</strong> response time <strong>of</strong> anesthesia trainees to simulated critical incidents. 8 They found<br />
great individual variability as well as variability by incident in the accuracy <strong>and</strong> timeliness <strong>of</strong><br />
response. Some simulated incidents, such as cardiac arrest, had major errors in management a<br />
majority <strong>of</strong> the time. Based on these studies, patient simulators can be used to identify areas for<br />
further education or training <strong>of</strong> anesthesia providers.<br />
We identified 2 studies <strong>of</strong> the effect <strong>of</strong> simulators on anesthetists’ performance. Schwid<br />
<strong>and</strong> colleagues studied the impact <strong>of</strong> a computer screen-based anesthesia simulator in a<br />
r<strong>and</strong>omized, controlled trial <strong>of</strong> 31 first-year anesthesia residents. 17 Residents that had trained on<br />
the simulator with individualized debriefing responded better to critical events on a mannequinbased<br />
simulator than those who received st<strong>and</strong>ard training without the simulator. Using a<br />
r<strong>and</strong>omized, controlled design, Chopra <strong>and</strong> colleagues studied management <strong>of</strong> simulated critical<br />
situations by 28 resident or staff anesthesiologists. 18 The performance <strong>of</strong> subjects who trained on<br />
the simulator was superior to that <strong>of</strong> subjects who did not receive that training.<br />
Another setting where simulators may play an important role is in crew resource<br />
management (CRM). Though establishing the effectiveness <strong>of</strong> simulation in CRM training may<br />
512
e difficult, 19 initial work has been done on reliably <strong>and</strong> consistently rating performance. 10 CRM<br />
is discussed further in Chapter 44.<br />
Pr<strong>of</strong>iciency on a simulator does not ensure pr<strong>of</strong>iciency in clinical settings. Simulator<br />
fidelity (ie, how accurately the simulator replicates reality) is imperfect. It is much more difficult<br />
to “re-create” a human being than to do so for, say, an airplane. This limitation is illustrated by a<br />
study conducted by Sayre <strong>and</strong> colleagues. They studied emergency medical technicians (EMT)<br />
who learned intubation techniques on anesthesia mannequins. 20 After successfully intubating the<br />
mannequins 10 times, they were permitted to intubate patients in the field, where their<br />
pr<strong>of</strong>iciency was only 53%. Other factors can inhibit optimum learning using simulation or the<br />
applicability <strong>of</strong> learning to real practice. Some participants may be more vigilant than usual<br />
during simulator sessions. Others may be unable to “suspend disbelief,” may treat the simulation<br />
only as a game, or act in a cavalier fashion, knowing that the simulator is not a real patient. 21<br />
Refinement <strong>of</strong> simulators to make them more sophisticated <strong>and</strong> life-like may help to improve the<br />
quality <strong>of</strong> the training that simulators can provide. Appropriate construction <strong>of</strong> curricula <strong>and</strong><br />
debriefings can also minimize the potential problems <strong>of</strong> simulation training.<br />
Simulators in Radiology<br />
The number <strong>of</strong> radiologic examinations that require sedation, analgesia, or contrast media<br />
has increased rapidly in recent years. 22 Despite their rarity, serious medication reactions do occur<br />
<strong>and</strong> require prompt, appropriate management. Some evidence <strong>of</strong> suboptimal management 23 has<br />
prompted the creation <strong>of</strong> computer-based simulators to improve training in these areas. 24<br />
Simulators have also been used to measure the effectiveness <strong>of</strong> strategies to teach trainees about<br />
critical incidents, but studies have not reported the effectiveness <strong>of</strong> simulators as adjuncts to<br />
training. Sica <strong>and</strong> colleagues developed a computer-based simulator that was used to study the<br />
effectiveness <strong>of</strong> a lecture <strong>and</strong> videotape-based intervention on critical incidents for radiology<br />
housestaff. 25 Those residents who underwent the intervention scored better on the simulator than<br />
those that had only received basic, st<strong>and</strong>ard training. The authors concluded that the simulator<br />
was an effective way <strong>of</strong> assessing the utility <strong>of</strong> the educational course.<br />
Simulators in Surgery<br />
As surgical technique <strong>and</strong> expertise has changed drastically over recent decades, some<br />
methods used to train surgeons have evolved as well. Simulators in the surgical setting are aimed<br />
at improving surgeons’ technical skills <strong>and</strong> dexterity. Training on simulators <strong>and</strong> virtual reality<br />
machines, though still in its nascent stages, is becoming increasingly accepted. 26,27 Surgical<br />
simulators have been developed for a variety <strong>of</strong> procedures: endovascular repair <strong>of</strong> abdominal<br />
aortic aneurysms, 28 sinus surgery 29,30 gynecologic surgery, 31 orthopedic surgery, 32 prostatic<br />
surgery, 33 amniocentesis procedures, 34 <strong>and</strong> oral surgery. 35 Nonetheless, many <strong>of</strong> these have yet to<br />
be formally evaluated in terms <strong>of</strong> efficacy in improving physician performance in patient care.<br />
We identified several studies that evaluated physician performance after training on a<br />
surgical simulator. Derossis evaluated surgical residents <strong>and</strong> attendings in a r<strong>and</strong>omized study<br />
<strong>and</strong> found that those trained with a simulator had greater pr<strong>of</strong>iciency in suturing, transferring,<br />
<strong>and</strong> mesh placement, when tested on the simulator, than did the control group. 36 They<br />
subsequently found that when tested in vivo in pigs, surgeons (both attendings <strong>and</strong> residents)<br />
who had been r<strong>and</strong>omized to the simulator arm were more pr<strong>of</strong>icient at the same skills. 37 Scott<br />
<strong>and</strong> colleagues studied the impact <strong>of</strong> a video-trainer on laparoscopic cholecystectomy skills. 38<br />
They r<strong>and</strong>omized surgical residents to training on a video-trainer versus no formal training over<br />
513
a 30-day period. At the end <strong>of</strong> the period, when tested on pre-specified tasks on the video trainer,<br />
those who trained on the video trainer did uniformly better.<br />
Intuitively, improved technical skills should lead to fewer complications during surgery.<br />
Wallwiener <strong>and</strong> colleagues developed a surgical trainer that, when used in conjunction with<br />
other improvements in their training program, led to lower rates <strong>of</strong> hysteroscopic<br />
complications. 39,40 However, for most simulators the link between improvements in technical<br />
skills <strong>and</strong> dexterity from simulator training <strong>and</strong> prevention <strong>of</strong> adverse events has yet to be<br />
established <strong>and</strong> deserves formal investigation. Further, problem-based surgical simulation (eg,<br />
avoiding inadvertent ligation <strong>of</strong> the ureter during hysterectomy) may improve patient safety not<br />
only by improving skills, but also by training surgeons to better anticipate <strong>and</strong> avoid<br />
complications <strong>and</strong> to manage them should they occur.<br />
Simulators in Gastroenterology<br />
Simulators have been developed to train physicians in the technical skills required in<br />
endoscopy. 41-43 We identified one study that evaluated the effect <strong>of</strong> simulator training on<br />
physician performance. In a small r<strong>and</strong>omized controlled trial enrolling 10 residents, Tuggy <strong>and</strong><br />
colleagues found residents who trained for flexible sigmoidoscopy using a virtual reality<br />
simulator were faster, visualized a greater portion <strong>of</strong> the colon, <strong>and</strong> made fewer directional errors<br />
in actual patients. 44<br />
Simulators in Cardiology<br />
Cardiology training has long used a variety <strong>of</strong> simulators from audiocassettes <strong>of</strong> heart<br />
tones to full patient simulators. Two simulators have been evaluated. Champagne <strong>and</strong> colleagues<br />
demonstrated that a heart sound simulator could increase medical students’ recognition <strong>of</strong><br />
pathologic heart sounds. 45 Ewy <strong>and</strong> colleagues studied the efficacy <strong>of</strong> “Harvey,” a cardiology<br />
patient simulator. 46 In a study enrolling 208 senior medical students at 5 medical schools,<br />
participants were r<strong>and</strong>omized to receive training on Harvey versus a st<strong>and</strong>ard cardiology<br />
curriculum during their cardiology elective. Students who had been trained with Harvey<br />
performed skills better both when tested with the simulator <strong>and</strong> when tested with real patients.<br />
Some physicians have expressed concern that training on simulators may decrease<br />
pr<strong>of</strong>essionalism. In this study, there was no difference in the way patients perceived the<br />
pr<strong>of</strong>essionalism <strong>of</strong> the students trained on Harvey compared with students who received st<strong>and</strong>ard<br />
training.<br />
Systems that simulate the cardiovascular anatomy <strong>and</strong> physiology have also been<br />
developed. Swanson <strong>and</strong> colleagues created a cardiovascular simulator to train physicians on the<br />
workings <strong>of</strong> mechanical valves <strong>and</strong> balloon assist devices, <strong>and</strong> to recognized diseased vessels. 47<br />
As cardiac procedures have become more invasive, simulators to train cardiologist in these<br />
procedures have become more common, 48, 49 but their effect on resident performance has not<br />
been evaluated formally.<br />
Comment<br />
Although simulators have been used for many years in a variety <strong>of</strong> settings, data on their<br />
efficacy are still emerging. While there is currently no evidence that simulation-based training<br />
leads to improved patient outcome, it may prove difficult to conduct such studies. These would<br />
require large cohorts <strong>of</strong> patients to be followed during <strong>and</strong> after care by clinicians who were<br />
r<strong>and</strong>omized to have undergone different cumulative amounts <strong>of</strong> simulation training. Because<br />
adverse events are uncommon, <strong>and</strong> there are a large number <strong>of</strong> patient-based <strong>and</strong> system-based<br />
514
factors that contribute to negative outcomes, any such study would have to be massive <strong>and</strong><br />
prolonged. Instead, provider performance, with its known limitations, has been <strong>and</strong> will continue<br />
to be used as a surrogate outcome. Nonetheless, as Gaba has asserted “…no industry in which<br />
human lives depend on skilled performance has waited for unequivocal pro<strong>of</strong> <strong>of</strong> the benefits <strong>of</strong><br />
simulation before embracing it.” 50 Certain benefits are clear. In training for procedures,<br />
simulators have high face validity because they ease trainees’ transition to actual patients, which<br />
seems inherently beneficial as a means to avoid adverse events. Further, procedural success is<br />
related to the experience <strong>of</strong> the operator, known as the volume-outcome relationship 51, 52 (see<br />
Chapter 19). As simulators become more advanced, they may be reasonable substitutes to<br />
improve pr<strong>of</strong>iciency <strong>of</strong> both trainees <strong>and</strong> low volume physician operators. This increase in<br />
pr<strong>of</strong>iciency may have an important impact in patient outcomes. Future studies <strong>of</strong> the link<br />
between simulator-based training <strong>and</strong> performance on actual patients will improve our ability to<br />
better assess the appropriate role <strong>of</strong> simulators in training <strong>and</strong> patient safety.<br />
The costs <strong>of</strong> simulators vary widely <strong>and</strong> need to be considered. “Home-made” or simple<br />
trainers are far less expensive than complex simulators or full-scale simulation centers. The<br />
average cost <strong>of</strong> high-fidelity patient simulators is on the order <strong>of</strong> $200,000. Medium fidelity<br />
simulators may be as little as $25,000. Establishing a dedicated simulation center can cost up to<br />
$1,000,000 (including the simulator) depending on the amount <strong>of</strong> space, the type <strong>of</strong> clinical<br />
equipment to be used, the extent <strong>of</strong> renovations needed, <strong>and</strong> the sophistication <strong>of</strong> the audiovisual<br />
equipment. However, such capital costs are amortized over a long period <strong>of</strong> time, <strong>and</strong> such<br />
centers typically are used for a wide variety <strong>of</strong> training curricula for diverse target populations.<br />
Further, for most simulation training the dominant cost is that <strong>of</strong> instructor time. Another indirect<br />
cost is that <strong>of</strong> removing clinical personnel from revenue producing work to undergo training.<br />
The health care industry currently does not fully embed time or costs <strong>of</strong> training into the system,<br />
but instead <strong>of</strong>ten leaves these costs for the individual clinicians to bear.<br />
There are potential risks to simulation-based training. Where the simulator cannot<br />
properly replicate the tasks or task environment <strong>of</strong> caring for patients, there is a risk that<br />
clinicians might acquire inappropriate behaviors (negative training) or develop a false sense <strong>of</strong><br />
security in their skills that could theoretically lead to harm. Although there are no data to suggest<br />
that this currently happens, such risks will have to be weighed <strong>and</strong> evaluated as simulators<br />
become more commonly used.<br />
In summary, although there is currently little evidence that simulation training improves<br />
patient care, the experience with simulation in other industries <strong>and</strong> the high face validity <strong>of</strong> their<br />
applications in health care has led many institutions to adopt the technology. It is likely that<br />
simulators will continue to be used <strong>and</strong> their role in training <strong>of</strong> medical personnel will grow.<br />
Definitive experiments to improve our underst<strong>and</strong>ing <strong>of</strong> their effects on training will allow them<br />
to be used more intelligently to improve provider performance, reduce errors <strong>and</strong> ultimately,<br />
promote patient safety. Although such experiments will be difficult <strong>and</strong> costly, they may be<br />
justified to determine how this technology can best be applied.<br />
References<br />
1. Bushell E, Gaba, D.M. Anesthesia Simulation <strong>and</strong> <strong>Patient</strong> <strong>Safety</strong>. Problems in Anesthesia.<br />
2001;In Press.<br />
2. Peugnet F, Dubois P, Roul<strong>and</strong> JF. Virtual reality versus conventional training in retinal<br />
photocoagulation: a first clinical assessment. Comput Aided Surg. 1998;3:20-26.<br />
3. Gaba DM, DeAnda A. A comprehensive anesthesia simulation environment: re-creating<br />
the operating room for research <strong>and</strong> training. Anesthesiology. 1988;69:387-394.<br />
515
4. Dunn D. Malignant hyperthermia. AORN J. 1997;65:728,731,passim.<br />
5. Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study<br />
<strong>of</strong> human factors. Anesthesiology. 1978;49:399-406.<br />
6. Spencer FC. Human error in hospitals <strong>and</strong> industrial accidents: current concepts. J Am Coll<br />
Surg. 2000;191:410-418.<br />
7. Keyser EJ, Derossis AM, Antoniuk M, Sigman HH, Fried GM. A simplified simulator for<br />
the training <strong>and</strong> evaluation <strong>of</strong> laparoscopic skills. Surg Endosc. 2000;14:149-153.<br />
8. Gaba DM, DeAnda A. The response <strong>of</strong> anesthesia trainees to simulated critical incidents.<br />
Anesth Analg. 1989;68:444-451.<br />
9. Gaba DM. Anaesthesia simulators. Can J Anaesth. 1995;42(10):952-953.<br />
10. Gaba DM, Howard SK, Flanagan B, Smith BE, Fish KJ, Botney R. Assessment <strong>of</strong> clinical<br />
performance during simulated crises using both technical <strong>and</strong> behavioral ratings.<br />
Anesthesiology. 1998;89:8-18.<br />
11. Ruppert M, Reith MW, Widmann JH, et al. Checking for breathing: evaluation <strong>of</strong> the<br />
diagnostic capability <strong>of</strong> emergency medical services personnel, physicians, medical<br />
students, <strong>and</strong> medical laypersons. Ann Emerg Med. 1999;34:720-729.<br />
12. Dorfsman ML, Menegazzi JJ, Wadas RJ, Auble TE. Two-thumb vs. two-finger chest<br />
compression in an infant model <strong>of</strong> prolonged cardiopulmonary resuscitation. Acad Emerg<br />
Med. 2000;7(10):1077-1082.<br />
13. Cheung V, Critchley LA, Hazlett C, Wong EL, Oh TE. A survey <strong>of</strong> undergraduate teaching<br />
in anaesthesia. Anaesthesia. 1999;54:4-12.<br />
14. King PH, Blanks ST, Rummel DM, Patterson D. Simulator training in anesthesiology: an<br />
answer? Biomed Instrum Technol. 1996;30:341-345.<br />
15. DeAnda A, Gaba DM. Unplanned incidents during comprehensive anesthesia simulation.<br />
Anesth Analg. 1990;71:77-82.<br />
16. Schwid HA, O'Donnell D. Anesthesiologists' management <strong>of</strong> simulated critical incidents.<br />
Anesthesiology. 1992;76:495-501.<br />
17. Schwid HA, Rooke GA, Michalowski P, Ross BK. Screen-based anesthesia simulation<br />
with debriefing improves performance in a mannequin-based anesthesia simulator. Teach<br />
Learn Med. 2001;13(2):92-96.<br />
18. Chopra V, Gesink BJ, de Jong J, Bovill JG, Spierdijk J, Br<strong>and</strong> R. Does training on an<br />
anaesthesia simulator lead to improvement in performance? Br J Anaesth. 1994;73:293-<br />
297.<br />
19. Gaba DM, Howard,S.K., Fish,K.J., Smith,B.E., Sowb,Y.A. Simulation-based training in<br />
anesthesia crisis resource management (ACRM): a decade <strong>of</strong> experience. Simulation <strong>and</strong><br />
Gaming. 2001;32(2):175-193.<br />
20. Sayre MR, Sakles JC, Mistler AF, Evans JL, Kramer AT, Pancioli AM. Field trial <strong>of</strong><br />
endotracheal intubation by basic EMTs. Ann Emerg Med. 1998;31:228-233.<br />
21. New Technologies in anesthesia practice: anesthesia simulators [Audiocassette]. Kansas<br />
City, MO: Audio-Digest Anesthesiology; 1998.<br />
22. Chait P. Future directions in interventional pediatric radiology. Pediatr Clin North Am.<br />
1997;44(3):763-782.<br />
23. Sadler DJ, Parrish F, Coulthard A. Intravenous contrast media reactions: how do<br />
radiologists react? Clin Radiol. 1994;49:879-882.<br />
24. Medina LS, Racadio JM, Schwid HA. Computers in radiology. The sedation, analgesia,<br />
<strong>and</strong> contrast media computerized simulator: a new approach to train <strong>and</strong> evaluate<br />
radiologists' responses to critical incidents. Pediatr Radiol. 2000;30:299-305.<br />
516
25. Sica GT, Barron DM, Blum R, Frenna TH, Raemer DB. Computerized realistic simulation:<br />
a teaching module for crisis management in radiology. AJR Am J Roentgenol.<br />
1999;172(2):301-304.<br />
26. Carrico CJ, Satava,R.M. Advanced simulation technologies for surgical education. Bull Am<br />
Coll Surg. 1996;81:77.<br />
27. Ota D, L<strong>of</strong>tin B, Saito T, Lea R, Keller J. Virtual reality in surgical education. Comput Biol<br />
Med. 1995;25:127-137.<br />
28. Chong CK, How TV, Black RA, Shortl<strong>and</strong> AP, Harris PL. Development <strong>of</strong> a simulator for<br />
endovascular repair <strong>of</strong> abdominal aortic aneurysms. Ann Biomed Eng. 1998;26:798-802.<br />
29. Rudman DT, Stredney D, Sessanna D, et al. Functional endoscopic sinus surgery training<br />
simulator. Laryngoscope. 1998;108(11 Pt 1):1643-1647.<br />
30. Ecke U, Klimek L, Muller W, Ziegler R, Mann W. Virtual reality: preparation <strong>and</strong><br />
execution <strong>of</strong> sinus surgery. Comput Aided Surg. 1998;3:45-50.<br />
31. Szekely G, Bajka M, Brechbuhler C, et al. Virtual reality based surgery simulation for<br />
endoscopic gynaecology. Stud Health Technol Inform. 1999;62:351-357.<br />
32. Poss R, Mabrey JD, Gillogly SD, et al. Development <strong>of</strong> a virtual reality arthroscopic knee<br />
simulator. J Bone Joint Surg Am. 2000;82-A(10):1495-1499.<br />
33. Ballaro A, Briggs T, Garcia-Montes F, MacDonald D, Emberton M, Mundy AR. A<br />
computer generated interactive transurethral prostatic resection simulator. J Urol.<br />
1999;162:1633-1635.<br />
34. Maher JE, Kleinman GE, Lile W, Tolaymat L, Steele D, Bernard J. The construction <strong>and</strong><br />
utility <strong>of</strong> an amniocentesis trainer. Am J Obstet Gynecol. 1998;179:1225-1227.<br />
35. Bettega G, Payan Y, Mollard B, Boyer A, Raphael B, Lavallee S. A simulator for<br />
maxill<strong>of</strong>acial surgery integrating 3D cephalometry <strong>and</strong> orthodontia. Comput Aided Surg.<br />
2000;5:156-165.<br />
36. Derossis AM, Bothwell J, Sigman HH, Fried GM. The effect <strong>of</strong> practice on performance in<br />
a laparoscopic simulator. Surg Endosc. 1998;12:1117-1120.<br />
37. Fried GM, Derossis AM, Bothwell J, Sigman HH. Comparison <strong>of</strong> laparoscopic<br />
performance in vivo with performance measured in a laparoscopic simulator. Surg Endosc.<br />
1999;13:1077-1081; discussion 1082.<br />
38. Scott DJ, Bergen PC, Rege RV, et al. Laparoscopic training on bench models: better <strong>and</strong><br />
more cost effective than operating room experience? J Am Coll Surg. 2000;191:272-283.<br />
39. Wallwiener D, Rimbach S, Bastert G. The HysteroTrainer, a simulator for diagnostic <strong>and</strong><br />
operative hysteroscopy. J Am Assoc Gynecol Laparosc. 1994;2:61-63.<br />
40. Wallwiener D, Rimbach S, Aydeniz B, Pollmann D, Bastert G. Operative Hysteroscopy:<br />
Results, Security Aspects, In Vitro Simulation Training (Hysterotrainer). J Am Assoc<br />
Gynecol Laparosc. 1994;1(4, Part 2):S39.<br />
41. Williams CB, Saunders BP, Bladen JS. Development <strong>of</strong> colonoscopy teaching simulation.<br />
Endoscopy. 2000;32:901-905.<br />
42. Aabakken L, Adamsen S, Kruse A. Performance <strong>of</strong> a colonoscopy simulator: experience<br />
from a h<strong>and</strong>s-on endoscopy course. Endoscopy. 2000;32:911-913.<br />
43. Neumann M, Mayer G, Ell C, et al. The Erlangen Endo-Trainer: life-like simulation for<br />
diagnostic <strong>and</strong> interventional endoscopic retrograde cholangiography. Endoscopy.<br />
2000;32(11):906-910.<br />
44. Tuggy ML. Virtual reality flexible sigmoidoscopy simulator training: impact on resident<br />
performance. J Am Board Fam Pract. 1998;11:426-433.<br />
517
45. Champagne MT, Harrell JS, Friedman BJ. Use <strong>of</strong> a heart sound simulator in teaching<br />
cardiac auscultation. Focus Crit Care. 1989;16:448-456.<br />
46. Ewy GA, Felner JM, Juul D, Mayer JW, Sajid AW, Waugh RA. Test <strong>of</strong> a cardiology<br />
patient simulator with students in fourth-year electives. J Med Educ. 1987;62:738-743.<br />
47. Swanson WM, Clark RE. A simple cardiovascular system simulator: design <strong>and</strong><br />
performance. J Bioeng. 1977;1:135-145.<br />
48. Wang Y, Chui C, Lim H, Cai Y, Mak K. Real-time interactive simulator for percutaneous<br />
coronary revascularization procedures. Comput Aided Surg. 1998;3:211-227.<br />
49. Dawson SL, Cotin S, Meglan D, Shaffer DW, Ferrell MA. Designing a computer-based<br />
simulator for interventional cardiology training. Catheter Cardiovasc Interv. 2000;51:522-<br />
527.<br />
50. Gaba DM. Improving anesthesiologists' performance by simulating reality. Anesthesiology.<br />
1992;76:491-494.<br />
51. Jollis JG, Romano PS. Volume-outcome relationship in acute myocardial infarction: the<br />
balloon <strong>and</strong> the needle. JAMA. 2000;284:3169-3171.<br />
52. Jollis JG, Peterson ED, Nelson CL, et al. Relationship between physician <strong>and</strong> hospital<br />
coronary angioplasty volume <strong>and</strong> outcome in elderly patients. Circulation. 1997;95:2485-<br />
2491.<br />
518
Chapter 46. Fatigue, Sleepiness, <strong>and</strong> <strong>Medical</strong> Errors<br />
Ashish K. Jha, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Bradford W. Duncan, MD<br />
Stanford University School <strong>of</strong> Medicine<br />
David W. Bates, MD, MSc<br />
Harvard <strong>Medical</strong> School<br />
Introduction<br />
Fatigue may contribute to the human error component <strong>of</strong> medical errors. 1-3 Hospitals<br />
function around the clock, which necessitates shift work for many personnel. Physicians,<br />
especially those in training, typically work long hours <strong>and</strong> are <strong>of</strong>ten sleep deprived. 4 Personnel<br />
who work during evenings <strong>and</strong> at night experience disruptions in circadian rhythms, which may<br />
aggravate fatigue. Although little research has focused specifically on fatigue in hospital<br />
personnel <strong>and</strong> its relationship to medical error, studies outside the medical field demonstrate the<br />
intuitive link between fatigue <strong>and</strong> degradation in performance <strong>and</strong> suggest some safety practices<br />
that may be adopted in medicine. Although both acute <strong>and</strong> chronic fatigue may have detrimental<br />
effects on the health <strong>of</strong> medical practitioners, 5-7 this chapter focuses on fatigue’s direct effects on<br />
patient safety. We review the literature on problem sleepiness among medical personnel, its<br />
impact on performance, <strong>and</strong> interventions to address sleep deprivation: limiting work hours,<br />
changes in shift scheduling, napping, <strong>and</strong> pharmaceutical aids. Although beyond the scope <strong>of</strong><br />
this chapter, factors that contribute to fatigue beyond sleepiness, such as job stress <strong>and</strong> work<br />
load, should be considered as part <strong>of</strong> a multifaceted strategy to combat fatigue.<br />
Background<br />
Fatigue <strong>and</strong> sleepiness may affect patient safety in several ways. Physicians <strong>and</strong> nurses<br />
need good attention, sound judgment, <strong>and</strong> <strong>of</strong>ten quick reaction time, especially in emergency<br />
situations. Whether evaluating an electrocardiogram for signs <strong>of</strong> myocardial ischemia or<br />
monitoring a patient during general anesthesia, degradation <strong>of</strong> attention, memory, or<br />
coordination may affect performance <strong>and</strong> lead to adverse events. Research suggests that sleep<br />
requirements <strong>and</strong> patterns are idiosyncratic, with wide variation across populations. In order to<br />
design interventions that will effectively decrease or prevent these events, it is important to<br />
underst<strong>and</strong> the signs, prevalence, <strong>and</strong> impact <strong>of</strong> sleep deprivation <strong>and</strong> problem sleepiness.<br />
Sleep Deprivation<br />
Individuals differ in their optimal sleep requirements. Most sleep experts agree that<br />
adults typically need between 6 <strong>and</strong> 10 hours <strong>of</strong> sleep per 24-hour period, with most people<br />
requiring approximately 8 hours <strong>of</strong> sleep per day. 8,9 When adults get less than 5 hours <strong>of</strong> sleep<br />
over a 24-hour period, peak mental abilities begin to decline. 2 For short periods <strong>of</strong> time (2-3<br />
days), adult who get 4 hours <strong>of</strong> sleep can function reasonably well, but below peak levels. 2<br />
However, even with sleep deprivation <strong>of</strong> just a couple <strong>of</strong> days, slower response times <strong>and</strong><br />
decreased initiatives are observed. 10 After one night <strong>of</strong> missed sleep, cognitive performance may<br />
decrease 25% from baseline. 11,12 After the second night <strong>of</strong> missed sleep, cognitive performance<br />
can fall to nearly 40% <strong>of</strong> baseline. 12<br />
With ongoing sleep deprivation (getting 2 to 3 hours less sleep than optimal), people<br />
develop a sleep debt. 2 If the sleep debt continues over 5 to 10 days, they are rarely maximally<br />
519
alert <strong>and</strong> at some point general performance, <strong>and</strong> particularly cognitive performance, become<br />
verifiably worse. Sleep debt also leads to slower response times, altered mood <strong>and</strong> motivation,<br />
<strong>and</strong> reduced morale <strong>and</strong> initiative. A meta-analysis <strong>of</strong> the effect <strong>of</strong> sleep deprivation on<br />
performance by Pilcher et al found that humans who are chronically sleep deprived function at<br />
the 9 th percentile <strong>of</strong> non-sleep-deprived subjects. Further, sleep deprivation affected mood more<br />
than it did cognitive function; both were more affected than motor function. 9<br />
Night-Shifts <strong>and</strong> Shift Rotation<br />
Shift work usually refers to a schedule in which some employees begin work at times<br />
other than the morning. In hospitals, up to 35% <strong>of</strong> nurses may be required to work at times other<br />
than the day shift. 13 A report by the Association <strong>of</strong> Pr<strong>of</strong>essional Sleep Societies concluded that<br />
night-time operators’ fatigue contributed to 4 well known disasters: Exxon Valdez, Bhopal,<br />
Chernobyl, <strong>and</strong> Three Mile Isl<strong>and</strong>. 14 Fatigue has also been implicated in aircraft accidents 15 <strong>and</strong><br />
in poor driving <strong>and</strong> accidents among truck drivers. 16 It is well documented that shift workers<br />
have disturbances in their circadian rhythm, as measured by changes in their melatonin <strong>and</strong><br />
cortisol levels. 17 Sleep after night work tends to be shorter than sleep after day work, leading to<br />
greater cumulative sleep deprivation. 18-20 Shift workers have poorer quality <strong>of</strong> sleep, marked by<br />
less REM sleep, <strong>and</strong> are less likely to feel refreshed after awaking. Between 60 <strong>and</strong> 70 percent <strong>of</strong><br />
shift workers complain <strong>of</strong> sleeping difficulties or problem sleepiness. 21 Several surveys <strong>of</strong> shift<br />
workers have found that those who work during night shifts are more likely to report sleepiness<br />
at work. 18,19,22,23 Alertness on the job is also affected, with employees showing less alertness<br />
during nighttime shifts. 24 In addition, shift workers tend to perform less well on reasoning <strong>and</strong><br />
non-stimulating tasks than non-shift workers. 22,23<br />
Prevalence <strong>and</strong> Severity<br />
Fatigue <strong>and</strong> sleep deprivation are common among medical personnel. Long work-hours<br />
are a tradition during residency, 25 with most interns <strong>and</strong> residents working 80 to 100 hours a<br />
week, <strong>of</strong>ten 36 hours at a time. 26 During these shifts their sleep is limited, <strong>and</strong> is usually<br />
interrupted. 27 In a 1991 national survey, second-year residents reported an average <strong>of</strong> 37.6 hours<br />
as the largest number <strong>of</strong> hours without sleep during their first postgraduate year <strong>and</strong> roughly<br />
25% <strong>of</strong> the residents reported being on call in the hospital over 80 hours per week. 26 A<br />
movement in the late 1980s, prompted partly by the death <strong>of</strong> a young woman, 28 led to regulations<br />
in New York State dictating that residents could work a maximum <strong>of</strong> 80 hours per week, with a<br />
maximum <strong>of</strong> 24 consecutive hours <strong>of</strong> patient care, <strong>and</strong> a minimum <strong>of</strong> 8 hours <strong>of</strong>f duty between<br />
shifts (see also chapter 55). 29 Despite these regulations, unannounced inspections <strong>of</strong> 12 teaching<br />
hospitals in New York State in March 1998 found 37% <strong>of</strong> all residents worked more than 85<br />
hours per week, 20% <strong>of</strong> all residents <strong>and</strong> 60% <strong>of</strong> surgical residents worked more than 95 hours<br />
per week, <strong>and</strong> 38% <strong>of</strong> all residents <strong>and</strong> 67% <strong>of</strong> all surgical residents worked more than 24<br />
consecutive hours. 30 In 2000, 8% <strong>of</strong> programs <strong>and</strong> institutions reviewed by the Accreditation<br />
Council for Graduate <strong>Medical</strong> Education were cited as being in violation <strong>of</strong> their work-hour<br />
requirements. 31 Work-hour violations were noted in general surgery (35%), pediatrics (16%),<br />
internal medicine (10%) <strong>and</strong> other training programs as well. 31<br />
Long hours <strong>and</strong> sleep deprivation continue after residency. Health care providers,<br />
particularly those still in training or who have recently completed training, occasionally work<br />
extra shifts to increase their income (“moonlighting”). One recent survey found that nearly half<br />
<strong>of</strong> all emergency medicine residents moonlight. 32 As many as 65% <strong>of</strong> internal medicine residents<br />
<strong>and</strong> fellows moonlight 33 <strong>and</strong> moonlighting is common among other residencies <strong>and</strong><br />
520
fellowships. 34, 35 These shifts are <strong>of</strong>ten at odd hours, <strong>and</strong> therefore are disruptive to normal sleep<br />
patterns. Among surgical staff, fatigue is common, especially since surgical teams can be<br />
involved in long, complicated operative cases that can take 12 to 20 hours at a time. 36,37<br />
Multiple studies have documented the impact <strong>of</strong> fatigue on medical personnel<br />
performance. 38 However, these studies have been limited by poor study designs or outcomes that<br />
may not correlate well with medical error. One study <strong>of</strong> nursing fatigue suggests that it may play<br />
a role in increased error. Gold <strong>and</strong> colleagues administered a questionnaire to nurses at a large<br />
academic hospital <strong>and</strong> found that nurses who worked a rotating schedule, when compared with<br />
nurses who predominantly worked day shifts, were more likely to fall asleep at work <strong>and</strong> get less<br />
sleep over all, <strong>and</strong> were nearly twice as likely to report committing a medication error. 39<br />
Using st<strong>and</strong>ardized testing, investigators have found that after a night <strong>of</strong> call, sleep<br />
deprived physicians may have worse language <strong>and</strong> numeric skills, 40 retention <strong>of</strong> information, 41<br />
short-term memory, 42 <strong>and</strong> concentration. 43 Performance on st<strong>and</strong>ardized tests may not reflect<br />
performance in medical situations. Taffinder et al studied the impact <strong>of</strong> sleep deprivation on<br />
surgical residents previously trained on a simulator <strong>and</strong> found that after a night without sleep,<br />
surgeons were slower <strong>and</strong> more prone to errors on the simulator than those who had a normal<br />
night <strong>of</strong> sleep. 44 Similarly, Denisco et al studied anesthesia residents after a night <strong>of</strong> sleep<br />
deprivation <strong>and</strong> found that those who had been on call <strong>and</strong> were sleep deprived scored less well<br />
on simulated critical events. 45 Smith-Coggins et al compared cognitive <strong>and</strong> motor performance<br />
<strong>of</strong> emergency physicians <strong>and</strong> found that, as the 24-hour study period progressed, physicians were<br />
more likely to make errors during a simulated triage test <strong>and</strong> while intubating a mannequin. 19<br />
However, other studies have failed to find an effect <strong>of</strong> sleep deprivation on cognitive<br />
performance by resident physicians. 46-48 Simulators may not reflect actual medical performance<br />
(see chapter 45). Though psychomotor performance seems to be affected by sleep deprivation,<br />
data are inconsistent as to fatigue’s impact on cognitive function <strong>and</strong> there are inadequate data<br />
assessing its impact on clinical performance.<br />
Few studies have looked at the impact <strong>of</strong> fatigue in hospital personnel on adverse events.<br />
A retrospective study by Haynes et al <strong>of</strong> 6371 surgical cases, found that the risk <strong>of</strong> postoperative<br />
complications among patients undergoing surgery was not increased when the surgical resident<br />
was sleep deprived. 49 These results may not be surprising for several reasons. First, the authors<br />
did not measure the residents’ error rate, which may have been higher with sleep deprivation.<br />
Second, the study did not measure the role attending physicians or other operating room<br />
personnel may have played in averting adverse events when residents erred. The supervisory<br />
aspect <strong>of</strong> system design can (<strong>and</strong> should) reduce both the frequency <strong>of</strong> individual mistakes (error<br />
prevention) <strong>and</strong> the likelihood <strong>of</strong> adverse events given that errors are inevitable (error<br />
absorption). 1 Finally, the rate <strong>of</strong> adverse events, including those that did not result in operative<br />
complications (“near misses”), may have been higher but under reported. Well-designed studies<br />
that evaluate the effects <strong>of</strong> fatigue among medical personnel on rates <strong>of</strong> medical errors or adverse<br />
events would be useful. In the meantime, the lack <strong>of</strong> convincing data linking fatigue with poor<br />
patient outcomes should not deter us from tackling the issue <strong>of</strong> fatigue among medical personnel.<br />
Practice Descriptions<br />
Hours <strong>of</strong> Service<br />
We reviewed the evidence for 2 potential safety practices concerning hours <strong>of</strong> service: 8hour<br />
versus 12-hour length shifts <strong>and</strong> regulations limiting maximum shift length <strong>and</strong>/or total<br />
hours worked. Most observational studies on optimal shift length to reduce fatigue <strong>and</strong> maximize<br />
521
performance are in non-medical settings <strong>and</strong> present inconsistent findings. In a study <strong>of</strong> workplace<br />
accidents in Germany, Hanecke et al found accident risk increased exponentially after the<br />
9 th hour at work <strong>and</strong> was highest among workers whose shift began in the evening or night. 50 The<br />
authors concluded that shifts that last longer than 8 hours might lead to more worker fatigue <strong>and</strong><br />
higher risk <strong>of</strong> accidents. Axelsson <strong>and</strong> colleagues studied workers at a power plant <strong>and</strong> found no<br />
difference in sleepiness or performance between those who worked 8-hour shifts <strong>and</strong> those who<br />
worked 12-hour shifts. 51 Another group found that switching from 8- to 12-hour shifts led to<br />
increased alertness on the job <strong>and</strong> improved recovery time after night shifts. 52 Overl<strong>and</strong> has<br />
proposed that work that requires complex cognitive tasks may be ill suited for longer shifts,<br />
whereas work with limited cognitive dem<strong>and</strong>s may be well suited for longer shifts. 53 Because the<br />
components <strong>of</strong> work vary dramatically within <strong>and</strong> across industries, shift durations that maintain<br />
performance in one setting may be ineffective in another.<br />
We identified 9 observational studies comparing 8- versus 12-hour shifts for medical<br />
personnel. Two studies <strong>of</strong> nursing care on 10 wards found that quantity 54 <strong>and</strong> quality 55 <strong>of</strong> care<br />
were significantly lower with 12-hour shifts. Six studies <strong>of</strong> nurses 56-61 <strong>and</strong> one <strong>of</strong> physicians 62<br />
measured outcomes including self-reported alertness, self-reported performance, <strong>and</strong>/or worker<br />
satisfaction. While 2 nurse studies found that self-reported alertness, performance, <strong>and</strong><br />
satisfaction wane with longer shifts, 56,57 Urgovics <strong>and</strong> Wright found that ICU nurses reported<br />
higher job satisfaction <strong>and</strong> subjectively improved clinical performance with 12-hour shifts. 60 The<br />
3 remaining studies in nurses found no difference in either satisfaction or self-reported<br />
performance between 8- <strong>and</strong> 12-hour shifts. 58,59,61 A survey <strong>of</strong> emergency department physicians<br />
found that those who worked 12-hour shifts were less likely to be satisfied than those who<br />
worked 8-hour shifts. 62 The relationship between these subjective outcomes measures <strong>and</strong><br />
medical error is not clear.<br />
Hours <strong>of</strong> service regulations as an effort to reduce errors due to fatigue are st<strong>and</strong>ard in<br />
some non-medical fields. Truck drivers are typically allowed to work no more than 10 hours at a<br />
time <strong>and</strong> no more than 60 hours in one week. Airline pilots <strong>and</strong> air traffic controllers work<br />
regulated hours <strong>and</strong> some data suggest waning performance as work-hours increase. 24,63-65<br />
Although most health care personnel are not subject to work-hour st<strong>and</strong>ards, many physiciansin-training<br />
are, either by statutory regulations or by being in an accredited training program. In a<br />
retrospective cohort study, Laine <strong>and</strong> colleagues found the aforementioned New York State<br />
regulations limiting resident work-hours had no effect on patient outcomes such as mortality or<br />
transfers to the intensive care unit but were associated with increased rates <strong>of</strong> medical<br />
complications <strong>and</strong> delays in diagnostic tests. 66 These negative effects may have been related to<br />
discontinuity <strong>of</strong> care <strong>and</strong>/or fewer physician-hours per patient. As the authors noted, “better care<br />
may be provided by a tired physician who is familiar with the patient than by a rested physician<br />
who is less familiar with the patient.” 66 In a case-control study, Petersen <strong>and</strong> colleagues found<br />
that when patients were cared for by a physician other than their primary resident, they were 6<br />
times as likely to suffer a preventable adverse event. 67 Thus, fewer physician work hours may<br />
lead to more physician discontinuity <strong>and</strong> potentially, more adverse events <strong>and</strong> poorer outcomes<br />
for patients.<br />
On the other h<strong>and</strong>, Gottlieb studied changes in a medical service staffing schedule that<br />
allowed for reduced sleep deprivation, improved distribution <strong>of</strong> admissions throughout the week,<br />
<strong>and</strong> improved continuity <strong>of</strong> inpatient care. 68 After these changes were instituted, patients had<br />
shorter lengths <strong>of</strong> stay, fewer ancillary tests, <strong>and</strong> fewer medication errors. Although it is difficult<br />
to ascribe the improvements to changes in work-hours because several other changes were made<br />
as well, it does appear that changes in work-hours can be made without adversely affecting<br />
522
patient outcomes. Any effort to change duty hours for health care personnel in an effort to reduce<br />
fatigue should factor in <strong>and</strong> continuously monitor numerous variables, including the potential<br />
costs <strong>of</strong> discontinuity, medical complications <strong>and</strong> unnecessary hospital days, to ensure that the<br />
measures do not compromise patient care. The costs needed to maintain adequate staffing in face<br />
<strong>of</strong> lost physician work-hours has been estimated to be $360 million in New York State alone. 69<br />
However, the difficult task <strong>of</strong> estimating other costs <strong>and</strong> potential savings from implementing<br />
these regulations has not been accomplished.<br />
Finally, some authors have expressed concern that restriction <strong>of</strong> resident physician workhours<br />
may lead to poorer quality training <strong>and</strong> decreased pr<strong>of</strong>essionalism among doctors. 70 They<br />
argue that restricted working hours will decrease a sense <strong>of</strong> obligation to patients <strong>and</strong> will<br />
sanction self-interest over the well-being <strong>of</strong> patients. However, there are no data to substantiate<br />
these concerns.<br />
Direction <strong>and</strong> Speed <strong>of</strong> Rotation <strong>of</strong> Shift Work<br />
The direction <strong>of</strong> shift rotation may impact worker fatigue. For workers who change from<br />
one shift to another, a forward rotation <strong>of</strong> shift work (morning shifts followed by evening shifts<br />
followed by night shifts) may lead to less fatigue on the job than backward rotation (day shift to<br />
night shift to evening shift). 71-74 Forward rotation appears easier to tolerate physiologically since<br />
the natural circadian rhythm tends to move forward <strong>and</strong> it is more difficult to fall asleep earlier<br />
than the normal bedtime. Several studies in non-medical personnel have shown that forward<br />
rotation allows for better acclimation <strong>of</strong> the circadian rhythm. 2,12,75 However, 2 other studies<br />
found no significant difference in forward versus backward shift rotation. 76,77 None <strong>of</strong> these<br />
studies measured worker performance or error rates <strong>and</strong> we found no studies that evaluated<br />
direction <strong>of</strong> shift work rotation among medical personnel.<br />
Another variable in scheduling is the speed <strong>of</strong> shift work rotation. Studies suggest that<br />
slow rotation (eg, changing from one shift to another every one to two weeks) may allow for<br />
better adaptation <strong>of</strong> the circadian rhythm than fast rotation (eg, changing shifts every 2-3<br />
days). 71,73,78,79 Slow shift rotation results in greater sleep length at home, less sleepiness on the<br />
job, better self-reported performance, <strong>and</strong> fewer errors. 74,79 In some cases, fast rotation may<br />
increase worker satisfaction 80 but the effects <strong>of</strong> such satisfaction on safety have not been<br />
assessed. Shift rotation at an extremely slow rate approximates fixed, non-rotating shifts<br />
(permanent night shifts, permanent day shifts). Permanent shifts are associated with better<br />
adaptation to changes in the circadian rhythm 78 <strong>and</strong> better performance than rotating shifts. 79<br />
However, daytime commitments <strong>and</strong> social obligations <strong>of</strong>ten prevent workers from completely<br />
adapting to permanent night shifts <strong>and</strong> worker satisfaction is poor. 71<br />
Improving Sleep: Education about Sleep Hygiene<br />
Good sleep hygiene, including the avoidance <strong>of</strong> alcohol <strong>and</strong> caffeine before bedtime, <strong>and</strong><br />
maintaining a healthy sleep environment, may aid in decreasing sleep debt <strong>and</strong> fatigue. Studies<br />
<strong>of</strong> sleep hygiene have focused on treatment <strong>of</strong> persons with insomnia or other chronic sleep<br />
disorders. 81-83 We found no clinical studies that measure the efficacy <strong>of</strong> good sleep hygiene<br />
among shift workers. Generally, most employers cannot dictate how their workers spend their<br />
hours <strong>of</strong>f-duty <strong>and</strong> compliance with recommendations may be poor. One study <strong>of</strong> lawenforcement<br />
<strong>of</strong>ficers working rotating shifts found significant increases in awareness <strong>and</strong><br />
knowledge after a training session on sleep hygiene practices but no change on a post-sleep<br />
523
inventory assessed at one-month follow-up. 84 The effectiveness <strong>of</strong> educational programs about<br />
sleep hygiene to improve shift worker performance requires further study.<br />
Lighting at Work<br />
The body’s regulation <strong>of</strong> circadian rhythm is mediated by the effects <strong>of</strong> light <strong>and</strong><br />
darkness. A 1986 survey found that 7.3 million Americans work at night. 71 These employees,<br />
who work during dark hours <strong>and</strong> sleep during daylight hours, are <strong>of</strong>ten chronically sleep<br />
deprived <strong>and</strong> may suffer adverse health effects, 85 partially due to poor synchrony <strong>of</strong> circadian<br />
rhythm to work schedule. Since scheduled light exposure can produce a phase shift in the<br />
endogenous circadian rhythm, 71, 86 investigators have studied changes in lighting at work <strong>and</strong><br />
home to improve adjustment to the shift cycle. Foret et al studied 8 young men in a sleep lab <strong>and</strong><br />
found exposure to bright lights during the night produced a beneficial effect on subjective<br />
alertness. 87 Czeisler <strong>and</strong> colleagues found that subjects who were exposed to bright light at night<br />
<strong>and</strong> nearly complete darkness during the day had better cognitive performance <strong>and</strong> subjective<br />
alertness, <strong>and</strong> longer daytime sleep (7.7 vs. 5.7 hours, p=0.01). 88<br />
Manipulation <strong>of</strong> light <strong>and</strong> dark is much easier in sleep labs than in the field, 89 where<br />
unintended exposure to bright light is common <strong>and</strong> may adversely impact attempts to alter<br />
workers’ circadian rhythm. 90 The National Aeronautics <strong>and</strong> Space Administration (NASA) has<br />
studied the efficacy <strong>of</strong> bright lights on shuttle astronauts. Their encouraging results suggest that<br />
alterations in circadian rhythm can be obtained upon on exposure to light at night. 89,91 The<br />
United States Nuclear Regulatory Commission has also implemented bright lighting for its night<br />
workers <strong>and</strong> found less fatigue <strong>and</strong> better alertness on the job. 92 Field studies are needed to<br />
determine how bright artificial light affects objective measures <strong>of</strong> performance in health care<br />
workers <strong>and</strong> medical error. Bright light may not be appropriate for all areas <strong>of</strong> the hospital. For<br />
example, Bullough <strong>and</strong> Rea have noted that while bright light might help workers in neonatal<br />
care units, it may also be detrimental to patients. 93<br />
Nonetheless, lighting can be a relatively inexpensive intervention using existing<br />
equipment. Keeping lights bright at night, <strong>and</strong> educating workers about using heavy shades at<br />
home may have an important impact on worker performance on night shifts.<br />
Napping<br />
Napping is common among shift workers <strong>and</strong> is perceived as a way to combat<br />
fatigue. 94,95 One study <strong>of</strong> shift workers in a steel plant found that over half reported napping at<br />
home either before or after their shifts. 94 The efficacy <strong>of</strong> naps has been studied in 3 settings:<br />
prior to periods <strong>of</strong> sleep deprivation (prophylactic naps), during periods <strong>of</strong> sleep deprivation<br />
(therapeutic naps) <strong>and</strong> during work hours (maintenance naps). Most studies have been<br />
conducted in sleep labs in healthy, young, male subjects.<br />
A number <strong>of</strong> studies in the non-medical literature have studied the efficacy <strong>of</strong><br />
prophylactic napping. Gillberg <strong>and</strong> colleagues studied 8 male subjects who were allowed only 4<br />
hours <strong>of</strong> sleep at night. When subjects took a 30-minute nap in the middle <strong>of</strong> the prior day, they<br />
had better subjective alertness, 20% improvement in vigilance performance, <strong>and</strong> less overall<br />
sleepiness than when they had not been allowed to nap. 96 Others have also found benefits <strong>of</strong><br />
prophylactic naps on subjective <strong>and</strong> objective measures <strong>of</strong> alertness <strong>and</strong> performance in healthy<br />
volunteers undergoing extended periods <strong>of</strong> sleep deprivation. 97-100 Bonnet <strong>and</strong> Ar<strong>and</strong> studied<br />
prophylactic versus therapeutic naps in 12 healthy young men who underwent 24 hours <strong>of</strong> sleep<br />
deprivation to simulate sleep patterns <strong>of</strong> medical housestaff. 101 One group <strong>of</strong> subjects had a 4-<br />
524
hour prophylactic nap in the evening <strong>and</strong> caffeine during the 24 hours, while the second group<br />
had four, 1-hour naps during the 24-hour work period <strong>and</strong> no caffeine. Those in the prophylactic<br />
nap <strong>and</strong> caffeine group had a 15% increase in reasoning <strong>and</strong> overall improved subjective<br />
alertness compared with the group that had only short naps. There was no impact on mood. We<br />
identified one study <strong>of</strong> napping by medical personnel. Harma <strong>and</strong> colleagues studied 146 female<br />
hospital nurses <strong>and</strong> nurses’ aides <strong>and</strong> found that those who napped prior to their night shifts were<br />
less likely to report on the job fatigue. 95<br />
Most studies evaluating the efficacy <strong>of</strong> therapeutic napping during prolonged periods <strong>of</strong><br />
sleep deprivation have found beneficial effects when compared with no napping. 102-108 On the<br />
other h<strong>and</strong>, Gillberg <strong>and</strong> colleagues found no difference in simulated driving between the 2<br />
groups <strong>of</strong> sleep deprived truck drivers, one group having taken a 30-minute nap during the<br />
middle <strong>of</strong> the previous night. 109<br />
Maintenance naps are naps that occur on the job, during the shift. These naps could<br />
compensate for daytime sleep deprivation or could bridge the nighttime low point in circadian<br />
somnolence. 110 Many Japanese industries have provided their employees with the option <strong>of</strong> on<br />
the job napping <strong>and</strong> nearly half <strong>of</strong> nighttime shift workers take advantage <strong>of</strong> this opportunity. 111<br />
Though no systematic studies <strong>of</strong> the impact <strong>of</strong> maintenance naps exist in shift workers, one<br />
investigation found that short naps in the middle <strong>of</strong> the night improved performance for the rest<br />
<strong>of</strong> the shift. 98 Napping over several successive shifts has not been studied. 110<br />
An important consideration in napping is the phenomena <strong>of</strong> sleep inertia, a period <strong>of</strong><br />
transitory hypo-vigilance, confusion, disorientation <strong>of</strong> behavior <strong>and</strong> impaired cognitive<br />
performance that immediately follows awakening. 112 Sleep inertia is well documented 112-116 <strong>and</strong><br />
lasts up to 30 minutes after awakening. 116-118 The duration <strong>of</strong> deep sleep <strong>and</strong> the time <strong>of</strong> the nap,<br />
relative to the circadian cycle, seem most related to the severity <strong>of</strong> sleep inertia. 8 Strategies for<br />
napping on the job to reduce fatigue should be designed to avoid possible detrimental effects <strong>of</strong><br />
sleep inertia. Another potential negative effect <strong>of</strong> lengthy naps is that they can disrupt the<br />
quantity <strong>and</strong> quality <strong>of</strong> later sleep periods. 119<br />
In summary, there is strong evidence that therapeutic naps <strong>and</strong> maintenance naps combat<br />
the effects <strong>of</strong> fatigue <strong>and</strong> sleep loss. They can help subjects adapt better to circadian rhythm<br />
disturbances <strong>and</strong> perform better during acute sleep deprivation. Their application in the medical<br />
field is not well known. While prophylactic <strong>and</strong> therapeutic napping result in loss <strong>of</strong> social time<br />
at home, maintenance napping results in loss <strong>of</strong> work time. Costs associated with naps have not<br />
been reported. The financial impact <strong>of</strong> reduced worker fatigue due to napping has not been<br />
evaluated in medicine.<br />
<strong>Medical</strong> Therapies<br />
Melatonin is the major hormone responsible for circadian rhythm regulation. James et al<br />
studied the effect <strong>of</strong> oral melatonin supplementation on circadian rhythm <strong>and</strong> adaptation to night<br />
shifts among medical personnel. 120 They <strong>and</strong> others have found no effect among medical shift<br />
workers. 121-123 Though melatonin continues to be studied for chronic insomnia <strong>and</strong> other<br />
conditions, there currently is insufficient evidence to recommend its use to combat the fatigue<br />
associated with changing workshifts.<br />
Some studies have looked at the potential benefits <strong>of</strong> benzodiazepines <strong>and</strong> other sedatives<br />
for short-term insomnia associated with shift work, but no data exist on long-term use.<br />
Stimulants <strong>and</strong> caffeine can boost performance acutely but do not address the underlying sleep<br />
deprivation, 124 <strong>and</strong> thus are not a viable long-term solution. Furthermore, concern over side<br />
525
effects, addiction, <strong>and</strong> performance degradation with current pharmacologic interventions makes<br />
their use as a safety practice unlikely.<br />
Comment<br />
Sleep deprivation <strong>and</strong> disturbances <strong>of</strong> circadian rhythm lead to fatigue, decreased<br />
alertness, <strong>and</strong> poor performance on st<strong>and</strong>ardized testing. Although data from non-medical fields<br />
suggest that sleep deprivation leads to poor job performance, this link has not yet been<br />
established in medicine. Although the link with fatigue seems intuitive, promoting interventions<br />
designed to combat medical errors should be evidence-based. Limits on physician duty hours<br />
must account for potentially detrimental effects <strong>of</strong> discontinuity in patient care. Forward rather<br />
than backward shift rotation, education about good sleep hygiene, <strong>and</strong> strategic napping before<br />
or during shifts may reduce fatigue <strong>and</strong> improve performance. High face validity, low likelihood<br />
<strong>of</strong> harm, <strong>and</strong> ease <strong>of</strong> implementation make these promising strategies, although more evidence <strong>of</strong><br />
their effectiveness in medicine is warranted. Studies on the use <strong>of</strong> bright light in the medical<br />
workplace are needed before it can be embraced.<br />
As Gaba points out, 125 in most high-hazard industries the assumption is that fatigue <strong>and</strong><br />
long, aberrant work hours lead to poor performance, <strong>and</strong> the burden <strong>of</strong> pro<strong>of</strong> is in the h<strong>and</strong>s <strong>of</strong><br />
those who believe that such work practices are safe. In medicine, concerns over discontinuity <strong>of</strong><br />
care, <strong>and</strong> difficulties in changing medical culture have pushed the burden <strong>of</strong> pro<strong>of</strong> into the h<strong>and</strong>s<br />
<strong>of</strong> those who wish to change the status quo. Given that medical personnel, like all human beings,<br />
probably function suboptimally when fatigued, efforts to reduce fatigue <strong>and</strong> sleepiness should be<br />
undertaken, <strong>and</strong> the burden <strong>of</strong> pro<strong>of</strong> should be in the h<strong>and</strong>s <strong>of</strong> the advocates <strong>of</strong> the current<br />
system to demonstrate that it is safe.<br />
Finally, fatigue among medical personnel may not be fully remediable <strong>and</strong> human errors<br />
are, in the end, inevitable. The ultimate solution for health care organizations will likely require a<br />
systems-based approach that both limits the potential for human error <strong>and</strong> intercepts errors that<br />
do occur before they reach patients.<br />
References<br />
1. Leape LL. Error in medicine. JAMA.1994; 272:1851-1857.<br />
2. Krueger GP. Fatigue, Performance, <strong>and</strong> <strong>Medical</strong> Error. In: Bogner MS, ed. Human Error in<br />
Medicine. Hillsdale, N.J.: L. Erlbaum Associates; 1994:311, 326.<br />
3. Kohn LT, Corrigan J, Donaldson MS. To err is human: building a safer health system.<br />
Washington, DC: National Academy Press; 2000.<br />
4. Lagnado L. Hospitals overwork young doctors in New York. Wall Street Journal. May 19,<br />
1998, 1998.<br />
5. Marcus CL, Loughlin GM. Effect <strong>of</strong> sleep deprivation on driving safety in housestaff.<br />
Sleep. 1996;19:763-766.<br />
6. Schwartz AJ, Black ER, Goldstein MG, Jozefowicz RF, Emmings FG. Levels <strong>and</strong> causes<br />
<strong>of</strong> stress among residents. J Med Educ. 1987;62:744-753.<br />
7. Small GW. House <strong>of</strong>ficer stress syndrome. Psychosomatics. 1981;22:860-869.<br />
8. Rosekind MR, G<strong>and</strong>er PH, Gregory KB, et al. Managing fatigue in operational settings 1:<br />
Physiological considerations <strong>and</strong> counter-measures. Hosp Top. 1997;75:23-30.<br />
9. Pilcher JJ, Huffcutt AI. Effects <strong>of</strong> sleep deprivation on performance: a meta-analysis.<br />
Sleep. 1996;19:318-326.<br />
10. Koslowsky M, Babk<strong>of</strong>f H. Meta-analysis <strong>of</strong> the relationship between total sleep<br />
deprivation <strong>and</strong> performance. Chronobiol Int. 1992;9:132-136.<br />
526
11. Cox T K, GP, ed. Stress <strong>and</strong> Sustained Performnace. 1 ed; 1989. Work <strong>and</strong> Stress; No. 3.<br />
12. Krueger G, ed. Sustained work, fatigue, sleep loss <strong>and</strong> performance: A review <strong>of</strong> the issues.<br />
1 ed; 1989. Work <strong>and</strong> Stress; No. 3.<br />
13. C<strong>of</strong>fey LC, Skipper JK, Jr., Jung FD. Nurses <strong>and</strong> shift work: effects on job performance<br />
<strong>and</strong> job-related stress. J Adv Nurs. 1988;13:245-254.<br />
14. Mitler MM, Carskadon MA, Czeisler CA, Dement WC, Dinges DF, Graeber RC.<br />
Catastrophes, sleep, <strong>and</strong> public policy: consensus report. Sleep. 1988;11:100-109.<br />
15. Rosekind MR, Gregory, K.B., Miller, D.L., Co EL, Lebacqz, J.V. Aircraft Accident<br />
Report: Uncontrolled Collision with Terrain, American International Airways Flight 808,<br />
Douglas DC-8, N814CK, U.S. Naval Air Station, Guantanamo Bay, Cuba, August 18,1993:<br />
National Transportation <strong>Safety</strong> Board; 1994.<br />
16. Wylie CD, Shultz, T., Miller, J.C. Commercial motor vehicle driver fatigue <strong>and</strong> alertness<br />
study: Essex Corporation; October 1996.<br />
17. Akerstedt T, Levi L. Circadian rhythms in the secretion <strong>of</strong> cortisol, adrenaline <strong>and</strong><br />
noradrenaline. Eur J Clin Invest. 1978;8:57-58.<br />
18. Salvendy G. H<strong>and</strong>book <strong>of</strong> human factors <strong>and</strong> ergonomics. 2nd ed. New York: Wiley; 1997.<br />
19. Smith-Coggins R, Rosekind MR, Hurd S, Buccino KR. Relationship <strong>of</strong> day versus night<br />
sleep to physician performance <strong>and</strong> mood. Ann Emerg Med. 1994;24:928-934.<br />
20. Smith-Coggins R, Rosekind MR, Buccino KR, Dinges DF, Moser RP. Rotating shiftwork<br />
schedules: can we enhance physician adaptation to night shifts? Acad Emerg Med.<br />
1997;4:951-961.<br />
21. Akerstedt T, Torsvall L. Shift work. Shift-dependent well-being <strong>and</strong> individual differences.<br />
Ergonomics. 1981;24:265-273.<br />
22. Akerstedt T, Kecklund G, Knutsson A. Manifest sleepiness <strong>and</strong> the spectral content <strong>of</strong> the<br />
EEG during shift work. Sleep. 1991;14:221-225.<br />
23. Akerstedt T. Sleepiness as a consequence <strong>of</strong> shift work. Sleep. 1988;11:17-34.<br />
24. Tucker P, Smith L, Macdonald I, Folkard S. Shift length as a determinant <strong>of</strong> retrospective<br />
on-shift alertness. Sc<strong>and</strong>inavian Journal <strong>of</strong> Work, Environment <strong>and</strong> Health. 1998;24 Suppl<br />
3:49-54.<br />
25. Green MJ. What (if anything) is wrong with residency overwork? Ann Intern Med.<br />
1995;123:512-517.<br />
26. Daugherty SR, Baldwin DC, Jr., Rowley BD. Learning, satisfaction, <strong>and</strong> mistreatment<br />
during medical internship: a national survey <strong>of</strong> working conditions. JAMA. 1998;279:1194-<br />
1199.<br />
27. Harvey R, Jarrett PG, Peltekian KM. Patterns <strong>of</strong> paging medical interns during night calls<br />
at two teaching hospitals. CMAJ. 1994;151:307-311.<br />
28. Asch DA, Parker RM. The Libby Zion case. One step forward or two steps backward? N<br />
Engl J Med. 1988;318:771-775.<br />
29. New York State Health Code. Section 405; 1989.<br />
30. DeBuono BA, Osten WM. The medical resident workload: the case <strong>of</strong> New York State.<br />
JAMA. 1998;280:1882-1883.<br />
31. Education ACfGM. Percent <strong>of</strong> programs cited for work hours 1999 <strong>and</strong> 2000:<br />
Accreditation Council for Graduate <strong>Medical</strong> Education; 2001 May 9 2001.<br />
32. Li J, Tabor R, Martinez M. Survey <strong>of</strong> moonlighting practices <strong>and</strong> work requirements <strong>of</strong><br />
emergency medicine residents. Am J Emerg Med. 2000;18:147-151.<br />
33. McCue JD, Janiszewski M, Stickley WT. Residents' views <strong>of</strong> the value <strong>of</strong> moonlighting.<br />
Arch Intern Med. 1990;150:1511-1513.<br />
527
34. Majidian AM, Brinker MR, Rice JC, Kerstein MD. Moonlighting in a university surgical<br />
training program. South Med J. 1993;86:441-446.<br />
35. Hunt KR, Hillman BJ, Witzke DB. Moonlighting during the radiology residency. Invest<br />
Radiol. 1992;27:978-983.<br />
36. Chalian AA, Kagan SH. Backside first in head <strong>and</strong> neck surgery?: preventing pressure<br />
ulcers in extended length surgeries. Head Neck. 2001;23:25-28.<br />
37. Greenberg MS. H<strong>and</strong>book <strong>of</strong> Neurosurgery. Vol 4. Lakel<strong>and</strong>, Florida: Greenberg Graphics,<br />
Inc.; 1997.<br />
38. Samk<strong>of</strong>f JS, Jacques CH. A review <strong>of</strong> studies concerning effects <strong>of</strong> sleep deprivation <strong>and</strong><br />
fatigue on residents' performance. Acad Med. 1991;66:687-693.<br />
39. Gold DR, Rogacz S, Bock N, et al. Rotating shift work, sleep, <strong>and</strong> accidents related to<br />
sleepiness in hospital nurses. Am J <strong>Public</strong> Health. 1992; 82:1011-1014.<br />
40. Hawkins MR, Vichick DA, Silsby HD, Kruzich DJ, Butler R. Sleep <strong>and</strong> nutritional<br />
deprivation <strong>and</strong> performance <strong>of</strong> house <strong>of</strong>ficers. J Med Educ. 1985;60:530-535.<br />
41. Hart RP, Buchsbaum DG, Wade JB, Hamer RM, Kwentus JA. Effect <strong>of</strong> sleep deprivation<br />
on first-year residents' response times, memory, <strong>and</strong> mood. J Med Educ. 1987;62:940-942.<br />
42. Rubin R, Orris P, Lau SL, Hryhorczuk DO, Furner S, Letz R. Neurobehavioral effects <strong>of</strong><br />
the on-call experience in housestaff physicians. J Occup Med. 1991;33:13-18.<br />
43. Robbins J, Gottlieb F. Sleep deprivation <strong>and</strong> cognitive testing in internal medicine house<br />
staff. West J Med. 1990;152:82-86.<br />
44. Taffinder NJ, McManus IC, Gul Y, Russell RC, Darzi A. Effect <strong>of</strong> sleep deprivation on<br />
surgeons' dexterity on laparoscopy simulator. Lancet. 1998;352:1191.<br />
45. Denisco RA, Drummond JN, Gravenstein JS. The effect <strong>of</strong> fatigue on the performance <strong>of</strong> a<br />
simulated anesthetic monitoring task. J Clin Monit. 1987;3:22-24.<br />
46. Reznick RK, Folse JR. Effect <strong>of</strong> sleep deprivation on the performance <strong>of</strong> surgical residents.<br />
Am J Surg. 1987;154:520-525.<br />
47. Deaconson TF, O'Hair DP, Levy MF, Lee MB, Schueneman AL, Codon RE. Sleep<br />
deprivation <strong>and</strong> resident performance. JAMA. 1988;260:1721-1727.<br />
48. Weinger MB, Vora, S. Hendon, C.N., Howard, S.K, Smith, B.E., Mazzei, W.J., Rosekind,<br />
M.R., Gaba D.M. Evaluation <strong>of</strong> the Effects <strong>of</strong> Fatigue <strong>and</strong> Sleepiness on Clinical<br />
Performance in On-call Anesthesia Residents During Actual Nighttime cases <strong>and</strong><br />
simulated cases. Paper presented at: Evaluating <strong>Patient</strong> <strong>Safety</strong> <strong>and</strong> Reducing Errors in<br />
Health Care; November 8-10, 1998, 1998; Rancho Mirage, CA.<br />
49. Haynes DF, Schwedler M, Dyslin DC, Rice JC, Kerstein MD. Are postoperative<br />
complications related to resident sleep deprivation? South Med J. 1995; 88:283-289.<br />
50. Hänecke K, Tiedemann S, Nachreiner F, Grzech-Sukalo H. Accident risk as a function <strong>of</strong><br />
hour at work <strong>and</strong> time <strong>of</strong> day as determined from accident data <strong>and</strong> exposure models for the<br />
German working population. Sc<strong>and</strong>inavian Journal <strong>of</strong> Work, Environment <strong>and</strong> Health.<br />
1998; 24 Suppl 3:43-48.<br />
51. Axelsson J, Kecklund G, Akerstedt T, Lowden A. Effects <strong>of</strong> alternating 8- <strong>and</strong> 12-hour<br />
shifts on sleep, sleepiness, physical effort <strong>and</strong> performance. Sc<strong>and</strong>inavian Journal <strong>of</strong> Work,<br />
Environment <strong>and</strong> Health. 1998; Vol 24 Suppl 3:62-68.<br />
52. Lowden A, Kecklund G, Axelsson J, Akerstedt T. Change from an 8-hour shift to a 12hour<br />
shift, attitudes, sleep, sleepiness <strong>and</strong> performance. Sc<strong>and</strong>inavian Journal <strong>of</strong> Work,<br />
Environment <strong>and</strong> Health. 1998; 24 Suppl 3:69-75.<br />
528
53. Overl<strong>and</strong> DW. Comparison <strong>of</strong> Effects <strong>of</strong> Change From 8 to 12 hour shifts on Air Force<br />
Aircraft Maintenance Workers. Alabama: Graduate School <strong>of</strong> Logistics Acquisition<br />
Management, Air Force Institute <strong>of</strong> Technology; 1997.<br />
54. Reid N, Robinson G, Todd C. The quantity <strong>of</strong> nursing care on wards working 8- <strong>and</strong> 12hour<br />
shifts. Int J Nurs Stud. 1993;30:403-413.<br />
55. Todd C, Reid N, Robinson G. The quality <strong>of</strong> nursing care on wards working eight <strong>and</strong><br />
twelve hour shifts: a repeated measures study using the MONITOR index <strong>of</strong> quality <strong>of</strong><br />
care. Int J Nurs Stud. 1989;26:359-368.<br />
56. Smith L, Folkard S, Tucker P, Macdonald I. Work shift duration: a review comparing eight<br />
hour <strong>and</strong> 12 hour shift systems. Occup Environ Med. 1998;55:217-229.<br />
57. Todd C, Robinson G, Reid N. 12-hour shifts: job satisfaction <strong>of</strong> nurses. J Nurs Manag.<br />
1993;1:215-220.<br />
58. Fields WL, Loveridge C. <strong>Critical</strong> thinking <strong>and</strong> fatigue: how do nurses on 8- & 12-hour<br />
shifts compare? Nurs Econ. 1988;6:189-195.<br />
59. Mills ME, Arnold B, Wood CM. Core-12: a controlled study <strong>of</strong> the impact <strong>of</strong> 12-hour<br />
scheduling. Nurs Res. 1983;32:356-361.<br />
60. Ugrovics A, Wright J. 12-hour shifts: does fatigue undermine ICU nursing judgments?<br />
Nurs Manage. 1990;21:64A, 64D, 64F-64G.<br />
61. Washburn MS. Fatigue <strong>and</strong> critical thinking on eight-<strong>and</strong> twelve-hour shifts. Nurs Manage.<br />
1991;22:80A, 80D, 80F-80H.<br />
62. Thomas H, Jr., Schwartz E, Whitehead DC. Eight- versus 12-hour shifts: implications for<br />
emergency physicians. Ann Emerg Med. 1994;23:1096-1100.<br />
63. Luna TD. Air traffic controller shiftwork: what are the implications for aviation safety? A<br />
review. Aviat Space Environ Med. 1997;68:69-79.<br />
64. Price WJ, Holley DC. Shiftwork <strong>and</strong> safety in aviation. Occup Med. 1990;5:343-377.<br />
65. Martin TC. Adverse effects <strong>of</strong> rotating schedules on the circadian rhythms <strong>of</strong> air medical<br />
crews. Air Med J. 1995;14:83-86.<br />
66. Laine C, Goldman L, Soukup JR, Hayes JG. The impact <strong>of</strong> a regulation restricting medical<br />
house staff working hours on the quality <strong>of</strong> patient care. JAMA. 1993; 269:374-378.<br />
67. Petersen LA, Brennan TA, O'Neil AC, Cook EF, Lee TH. Does housestaff discontinuity <strong>of</strong><br />
care increase the risk for preventable adverse events? Ann Intern Med. 1994; 121:866-872.<br />
68. Gottlieb DJ, Parenti CM, Peterson CA, L<strong>of</strong>gren RP. Effect <strong>of</strong> a change in house staff work<br />
schedule on resource utilization <strong>and</strong> patient care. Arch Intern Med. 1991; 151:2065-2070.<br />
69. Thorpe KE. House staff supervision <strong>and</strong> working hours. Implications <strong>of</strong> regulatory change<br />
in New York State. JAMA. 1990;263:3177-3181.<br />
70. Holzman IR, Barnett SH. The Bell Commission: ethical implications for the training <strong>of</strong><br />
physicians. Mt Sinai J Med. 2000;67:136-139.<br />
71. Czeisler CA, Moore-Ede MC, Coleman RH. Rotating shift work schedules that disrupt<br />
sleep are improved by applying circadian principles. Science. 1982;217:460-463.<br />
72. Akerstedt T. Shifted sleep hours. Ann Clin Res. 1985;17:273-279.<br />
73. Folkard S. Shiftwork <strong>and</strong> performance. Paper presented at: Symposium on the Variations<br />
<strong>of</strong> Work-Sleep Schedules, 1981.<br />
74. Knauth P. Speed <strong>and</strong> direction <strong>of</strong> shift rotation. J Sleep Res. 1995; 4:41-46.<br />
75. Comperatore C.A. K, G.P. Circadian rhythm desynchronosis, jet lag, shift lag, <strong>and</strong> coping<br />
strategies. In: A.J. S, ed. Occupational medicine: Shiftwork. State <strong>of</strong> the art reviews.<br />
Philadelphia, PA: Hanley & Belfus; 1990:323-341.<br />
529
76. Tucker P, Smith L, Macdonald I, Folkard S. Effects <strong>of</strong> direction <strong>of</strong> rotation in continuous<br />
<strong>and</strong> discontinuous 8 hour shift systems. Occup Environ Med. 2000;57:678-684.<br />
77. Barton J, Folkard S, Smith L, Poole CJ. Effects on health <strong>of</strong> a change from a delaying to an<br />
advancing shift system. Occup Environ Med. 1994;51:749-755.<br />
78. Knauth P, Rutenfranz J, Schulz H, et al. Experimental shift work studies <strong>of</strong> permanent<br />
night, <strong>and</strong> rapidly rotating, shift systems. II. Behaviour <strong>of</strong> various characteristics <strong>of</strong> sleep.<br />
Int Arch Occup Environ Health. 1980;46:111-125.<br />
79. Pilcher JJ, Lambert BJ, Huffcutt AI. Differential effects <strong>of</strong> permanent <strong>and</strong> rotating shifts on<br />
self-report sleep length: a meta-analytic review. Sleep. 2000;23:155-163.<br />
80. Williamson AM, S<strong>and</strong>erson JW. Changing the speed <strong>of</strong> shift rotation: a field study.<br />
Ergonomics. 1986;29:1085-1095.<br />
81. Morin CM, Colecchi C, Stone J, Sood R, Brink D. Behavioral <strong>and</strong> pharmacological<br />
therapies for late-life insomnia: a r<strong>and</strong>omized controlled trial. JAMA. 1999;281:991-999.<br />
82. Friedman L, Benson K, Noda A, et al. An actigraphic comparison <strong>of</strong> sleep restriction <strong>and</strong><br />
sleep hygiene treatments for insomnia in older adults. J Geriatr Psychiatry Neurol.<br />
2000;13:17-27.<br />
83. Vgontzas AN, Kales A. Sleep <strong>and</strong> its disorders. Annu Rev Med. 1999;50:387-400.<br />
84. Holbrook MI, White MH, Hutt MJ. Increasing awareness <strong>of</strong> sleep hygiene in rotating shift<br />
workers: arming law-enforcement <strong>of</strong>ficers against impaired performance. Percept Mot<br />
Skills. 1994;79:520-522.<br />
85. Monk TH, Folkard, S. Making Shift Work Tolerable. London: Taylor <strong>and</strong> Frances; 1992.<br />
86. Budnick LD, Lerman SE, Nicolich MJ. An evaluation <strong>of</strong> scheduled bright light <strong>and</strong><br />
darkness on rotating shiftworkers: trial <strong>and</strong> limitations. Am J Ind Med. 1995; 27:771-782.<br />
87. Foret J, Daurat A, Tirilly G. Effect <strong>of</strong> bright light at night on core temperature, subjective<br />
alertness <strong>and</strong> performance as a function <strong>of</strong> exposure time. Sc<strong>and</strong> J Work Environ Health.<br />
1998; 24:115-120.<br />
88. Czeisler CA, Johnson MP, Duffy JF, Brown EN, Ronda JM, Kronauer RE. Exposure to<br />
bright light <strong>and</strong> darkness to treat physiologic maladaptation to night work. N Engl J Med.<br />
1990;322:1253-1259.<br />
89. Eastman CI, Boulos Z, Terman M, Campbell SS, Dijk DJ, Lewy AJ. Light treatment for<br />
sleep disorders: consensus report. VI. Shift work. J Biol Rhythms. 1995;10:157-164.<br />
90. Chesson AL, Jr., Littner M, Davila D, et al. Practice parameters for the use <strong>of</strong> light therapy<br />
in the treatment <strong>of</strong> sleep disorders. St<strong>and</strong>ards <strong>of</strong> Practice Committee, American Academy<br />
<strong>of</strong> Sleep Medicine. Sleep. 1999;22:641-660.<br />
91. Putcha L DD, Nimmagudda R, Stewart KT, Kripke DF. Chronobiological measurements in<br />
astronauts. Sleep Research. 1997;26:746.<br />
92. Baker TL. Use <strong>of</strong> a circadian lighting system to improve night shift alertness <strong>and</strong><br />
performance at the USNRC's Headquarters Operations Center. Paper presented at: <strong>Safety</strong><br />
<strong>of</strong> Operating Nuclear Reactions; September 17-20, 1995, 2001; Seattle, WA.<br />
93. Bullough J, Rea, M.S. Lighting for neonatal intensive care units: some critical information<br />
for design. Lighting Research <strong>and</strong> Technology. 1996;28:189-198.<br />
94. Akerstedt T, Torsvall L. Napping in shift work. Sleep. 1985;8:105-109.<br />
95. Harma M, Knauth P, Ilmarinen J. Daytime napping <strong>and</strong> its effects on alertness <strong>and</strong> shortterm<br />
memory performance in shiftworkers. Int Arch Occup Environ Health. 1989;61:341-<br />
345.<br />
96. Gillberg M, Kecklund G, Axelsson J, Akerstedt T. The effects <strong>of</strong> a short daytime nap after<br />
restricted night sleep. Sleep. 1996;19:570-575.<br />
530
97. Nicholson AN, Pascoe PA, Roehrs T, et al. Sustained performance with short evening <strong>and</strong><br />
morning sleeps. Aviat Space Environ Med. 1985;56:105-114.<br />
98. Gillberg M. The effects <strong>of</strong> two alternative timings <strong>of</strong> a one-hour nap on early morning<br />
performance. Biol Psychol. 1984;19:45-54.<br />
99. Dinges DF, Orne MT, Whitehouse WG, Orne EC. Temporal placement <strong>of</strong> a nap for<br />
alertness: contributions <strong>of</strong> circadian phase <strong>and</strong> prior wakefulness. Sleep. 1987;10:313-329.<br />
100. Taub JM. Effects <strong>of</strong> scheduled afternoon naps <strong>and</strong> bedrest on daytime alertness. Int J<br />
Neurosci. 1982;16:107-127.<br />
101. Bonnet MH, Ar<strong>and</strong> DL. Impact <strong>of</strong> naps <strong>and</strong> caffeine on extended nocturnal performance.<br />
Physiol Behav. 1994;56:103-109.<br />
102. Taub JM, Tanguay PE, Clarkson D. Effects <strong>of</strong> daytime naps on performance <strong>and</strong> mood in a<br />
college student population. J Abnorm Psychol. 1976;85:210-217.<br />
103. Lubin A, Hord DJ, Tracy ML, Johnson LC. Effects <strong>of</strong> exercise, bedrest <strong>and</strong> napping on<br />
performance decrement during 40 hours. Psychophysiology. 1976;13:334-339.<br />
104. Lumley M, Roehrs T, Zorick F, Lamphere J, Roth T. The alerting effects <strong>of</strong> naps in sleepdeprived<br />
subjects. Psychophysiology. 1986;23:403-408.<br />
105. Mullaney DJ, Kripke DF, Fleck PA, Johnson LC. Sleep loss <strong>and</strong> nap effects on sustained<br />
continuous performance. Psychophysiology. 1983;20:643-651.<br />
106. Matsumoto K. Effects <strong>of</strong> nighttime naps on body temperature changes, sleep patterns, <strong>and</strong><br />
self-evaluation <strong>of</strong> sleep. J Hum Ergol (Tokyo). 1981;10:173-184.<br />
107. Naitoh P, Englund CE, Ryman D. Restorative power <strong>of</strong> naps in designing continuous work<br />
schedules. J Hum Ergol (Tokyo). 1982;11:259-278.<br />
108. Rosekind MR, Graeber, R. C., Dinges, D. F., Connell, L. J., Rountree, M. S., Spinweber C.<br />
L., Gillen, K. A. Crew Factors in Flight Operations IX: Effects <strong>of</strong> Planned Cockpit Rest on<br />
Crew Performance <strong>and</strong> Alertness in Long-Haul Operations. M<strong>of</strong>fett Field, California:<br />
NASA; 1994.<br />
109. Gillberg M, Kecklund G, Akerstedt T. Sleepiness <strong>and</strong> performance <strong>of</strong> pr<strong>of</strong>essional drivers<br />
in a truck simulator- -comparisons between day <strong>and</strong> night driving. J Sleep Res. 1996;5:12-<br />
15.<br />
110. Rosa RR, Bonnet MH, Bootzin RR, et al. Intervention factors for promoting adjustment to<br />
nightwork <strong>and</strong> shiftwork. Occup Med. 1990;5:391-415.<br />
111. Reinberg A, Andlauer P, Vieux N, Permanent Commission <strong>and</strong> International Association<br />
on Occupational Health. Scientific Committee on Shift Work. Night <strong>and</strong> shift work :<br />
biological <strong>and</strong> social aspects : proceedings <strong>of</strong> the Fifth International Symposium on Night<br />
<strong>and</strong> Shift Work : Scientific Committee on Shift Work <strong>of</strong> the Permanent Commission <strong>and</strong><br />
International Association on Occupational Health (PCIAOH) Rouen, 12-16 May 1980. 1st<br />
ed. Oxford ; New York: Pergamon Press; 1981.<br />
112. Ferrara M, De Gennaro L. The sleep inertia phenomenon during the sleep-wake transition:<br />
theoretical <strong>and</strong> operational issues. Aviat Space Environ Med. 2000;71:843-848.<br />
113. Takahashi M, Arito H, Fukuda H. Nurses' workload associated with 16-h night shifts. II:<br />
Effects <strong>of</strong> a nap taken during the shifts. Psychiatry Clin Neurosci. 1999;53:223-225.<br />
114. Ferrara M, De Gennaro L, Bertini M. The effects <strong>of</strong> slow-wave sleep (SWS) deprivation<br />
<strong>and</strong> time <strong>of</strong> night on behavioral performance upon awakening. Physiol Behav. 1999;68:55-<br />
61.<br />
115. Ferrara M, De Gennaro L, Casagr<strong>and</strong>e M, Bertini M. Selective slow-wave sleep<br />
deprivation <strong>and</strong> time-<strong>of</strong>-night effects on cognitive performance upon awakening.<br />
Psychophysiology. 2000;37:440-446.<br />
531
116. Bruck D, Pisani DL. The effects <strong>of</strong> sleep inertia on decision-making performance. J Sleep<br />
Res. 1999;8:95-103.<br />
117. Ferrara M, De Gennaro L, Bertini M. Time-course <strong>of</strong> sleep inertia upon awakening from<br />
nighttime sleep with different sleep homeostasis conditions. Aviat Space Environ Med.<br />
2000;71:225-229.<br />
118. Jewett ME, Wyatt JK, Ritz-De Cecco A, Khalsa SB, Dijk DJ, Czeisler CA. Time course <strong>of</strong><br />
sleep inertia dissipation in human performance <strong>and</strong> alertness. J Sleep Res. 1999;8:1-8.<br />
119. Rosekind MR, Smith RM, Miller DL, et al. Alertness management: strategic naps in<br />
operational settings. J Sleep Res. 1995;4:62-66.<br />
120. James M, Tremea MO, Jones JS, Krohmer JR. Can melatonin improve adaptation to night<br />
shift? Am J Emerg Med. 1998; 16:367-370.<br />
121. Jorgensen KM, Witting MD. Does exogenous melatonin improve day sleep or night<br />
alertness in emergency physicians working night shifts? Ann Emerg Med. 1998;31:699-<br />
704.<br />
122. Wrenn K, Wright S. Melatonin after night shift work. Ann Emerg Med. 1999;33:479.<br />
123. Wright SW, Lawrence LM, Wrenn KD, Haynes ML, Welch LW, Schlack HM.<br />
R<strong>and</strong>omized clinical trial <strong>of</strong> melatonin after night-shift work: efficacy <strong>and</strong><br />
neuropsychologic effects. Ann Emerg Med. 1998;32:334-340.<br />
124. Walsh JK, Muehlbach MJ, Humm TM, Dickins QS, Sugerman JL, Schweitzer PK. Effect<br />
<strong>of</strong> caffeine on physiological sleep tendency <strong>and</strong> ability to sustain wakefulness at night.<br />
Psychopharmacology. 1990;101:271-273.<br />
125. Gaba DM. Physician work hours: The "sore thumb" <strong>of</strong> organizational safety in tertiary<br />
health care. Paper presented at: Proceedings <strong>of</strong> Enhancing <strong>Patient</strong> <strong>Safety</strong> <strong>and</strong> Reducing<br />
Errors in Health Care, 1998; Rancho Mirage, CA.<br />
532
533
Chapter 47. <strong>Safety</strong> During Transport <strong>of</strong> <strong>Critical</strong>ly Ill <strong>Patient</strong>s<br />
Susana B. Martins, MD, MSc<br />
Kaveh G. Shojania, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
<strong>and</strong> Institute for Health Policy Studies<br />
Background<br />
The care <strong>of</strong> acutely ill patients routinely includes transportation, both within a given<br />
hospital to undergo tests <strong>and</strong> procedures, <strong>and</strong> between hospitals, as patients may require transfer<br />
to other facilities for specialized services. <strong>Critical</strong>ly ill patients in particular commonly require<br />
such transfers <strong>and</strong> are at high risk for complications en route. 1-4 Developing practices to reduce<br />
or minimize this necessary risk represents a potentially important area <strong>of</strong> patient safety research.<br />
This chapter focuses on transportation <strong>of</strong> critically ill patients by health pr<strong>of</strong>essionals<br />
(paramedics, nurses, physicians <strong>and</strong>/or respiratory therapists) between hospitals (to receive<br />
higher levels <strong>of</strong> care) <strong>and</strong> within the hospital (for diagnostic or therapeutic procedures).<br />
Stabilization before transport, in the field or in the transferring hospital, <strong>and</strong> the mode <strong>of</strong><br />
transferring patients from the field to specialized centers also present important research <strong>and</strong><br />
policy questions. 5 However, we regarded these issues as clinical research topics <strong>and</strong> quality<br />
improvement issues for the fields <strong>of</strong> pre-hospital <strong>and</strong> emergency medicine, rather than patient<br />
safety in general, <strong>and</strong> so do not review this literature here.<br />
Practice Description<br />
Intrahospital transport refers to transportation <strong>of</strong> patients within a hospital for the<br />
purpose <strong>of</strong> undergoing diagnostic or therapeutic procedures or transfer to a specialized unit. In<br />
the context <strong>of</strong> this chapter, this generally involves movement <strong>of</strong> critically ill patients from<br />
intensive care areas <strong>of</strong> the hospital (including intensive care units, emergency departments,<br />
operating theaters <strong>and</strong> recovery rooms) to areas typically not involved in the delivery <strong>of</strong> such<br />
care (eg, a hospital radiology department). Equipment <strong>and</strong> staffing used for intrahospital<br />
transport varies by hospital, clinical service <strong>and</strong> patient acuity. Studies <strong>of</strong> intrahospital transport<br />
have mainly focused on the adequacy <strong>of</strong> patient monitoring <strong>and</strong> ventilator support. The specific<br />
practices evaluated in this chapter include:<br />
• The continued use <strong>of</strong> mechanical ventilation instead <strong>of</strong> switching to manual<br />
ventilation. Manual ventilation involves a self-inflating bag with or without a<br />
volumeter, while mechanical ventilation consists <strong>of</strong> a portable, time-cycled,<br />
volume-constant transport ventilator.<br />
• The use <strong>of</strong> specialized transfer units during intrahospital transport. The unit is<br />
attached to the patient’s bed <strong>and</strong> contains all equipment necessary to meet the<br />
patient’s needs (ventilation, monitoring <strong>and</strong> infusion <strong>of</strong> drugs) in the ICU <strong>and</strong><br />
during transport. The unit works as a st<strong>and</strong>-alone unit.<br />
Interhospital transport refers to transportation <strong>of</strong> patients between hospitals by ground or<br />
air ambulance. Interhospital transport teams vary widely in composition, training <strong>and</strong> experience.<br />
The transport team does not always include a physician; even when a physician is present, his or<br />
her training may not include skills necessary for this task. 6-8 Nurses <strong>and</strong> respiratory therapists<br />
frequently accompany critically ill patients during interhospital transport. Some paramedics<br />
534
eceive special training in skills necessary for the interhospital transport <strong>of</strong> critically ill patients. 9<br />
As with physicians, the training <strong>of</strong> nurses <strong>and</strong> respiratory therapists assigned responsibility for<br />
interhospital transport varies widely. Equipment used during interhospital transport also varies<br />
widely, 6,7 but the practices evaluated in the literature mainly relate to the use <strong>of</strong> specialized<br />
transport teams.<br />
Specialized transport teams characteristically receive consistent <strong>and</strong> high levels <strong>of</strong><br />
training <strong>and</strong> experience in the transportation <strong>of</strong> critically ill patients, 10-12 compared with teams<br />
assembled ad hoc. Further details <strong>of</strong> the composition <strong>of</strong> these teams are presented in connection<br />
with the specific studies reviewed below (see Table 47.1). Because <strong>of</strong> the relative paucity <strong>of</strong><br />
studies <strong>of</strong> practices for improving the safety <strong>of</strong> patient transport, we have reviewed the pediatric<br />
<strong>and</strong> adult literature together.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Adverse events during transport <strong>of</strong> critically ill patients fall into two general categories:<br />
mishaps related to intensive care (eg, lead disconnections, loss <strong>of</strong> battery power, loss <strong>of</strong><br />
intravenous access, accidental extubation, occlusion <strong>of</strong> the endotracheal tube, or exhaustion <strong>of</strong><br />
oxygen supply), <strong>and</strong> physiologic deteriorations related to critical illness (eg, worsening<br />
hypotension or hypoxemia). Unfortunately many studies do not distinguish clearly between these<br />
2 categories. Further complicating assessments <strong>of</strong> patient transport as a safety problem is the<br />
confounding effect <strong>of</strong> patient selection, as patents requiring intra- or interhospital transport likely<br />
represent a sicker patient population than unselected critically ill patients. In fact, one casecontrol<br />
study reported no differences in adverse events (equipment-related or physiologic) in<br />
critically ill adults during the period <strong>of</strong> intrahospital transportation as compared to matched<br />
subjects in the ICU. 13<br />
Death during transport is a rare event. The majority <strong>of</strong> studies reported no mortality<br />
during intrahospital transport 13-17 or interhospital transport, 18,19 <strong>and</strong> some do not mention<br />
deaths.21-23,24<br />
For intrahospital transport <strong>of</strong> critically ill patients, reported rates <strong>of</strong> adverse events range<br />
from 5.9% to 66%. 13,14,16,21,22,25 (We could find no comparable reports <strong>of</strong> event rates for critically<br />
ill children.) Much <strong>of</strong> this variation undoubtedly reflects definitional differences, but differences<br />
in patient populations also contribute to this wide range. For instance, a prospective study <strong>of</strong> 50<br />
high-risk adult cardiac patients reported arrhythmias in 84% <strong>of</strong> patients, with 52% <strong>of</strong> these<br />
arrhythmias providing an indication for emergency treatment. 17 These event rates are clearly<br />
much higher than would be observed in an unselected population <strong>of</strong> critically patients. Similarly,<br />
Insel et al showed a significantly higher incidence <strong>of</strong> hemodynamic changes requiring<br />
therapeutic intervention when intrahospital transport involved transfers from the operating room<br />
to the ICU compared with patients transported from the ICU to diagnostic procedures. 26<br />
In contrast to the above, the literature on adverse events during interhospital transport has<br />
generally involved critically ill children, not adults. Reported rates <strong>of</strong> adverse events during<br />
pediatric interhospital transport range from 0 to 75%. 2,10-12,19,24,27-29 In one <strong>of</strong> these studies, a<br />
prospective cohort design reported a morbidity ratio <strong>of</strong> 1.85 (95% CI: 1.12-3.06) for pediatric<br />
patients transported from another hospital to the pediatric ICU (PICU) as compared with those<br />
admitted directly (emergency room <strong>and</strong> wards). Importantly, this increased morbidity reflected<br />
an increased rate <strong>of</strong> “intensive care events” such as plugged endotracheal tubes <strong>and</strong> loss <strong>of</strong><br />
intravenous access, not an increase in physiologic events. <strong>Patient</strong>s experiencing such adverse<br />
events tended to have higher morbidity scores (on the PRISM scale) <strong>and</strong> lower therapy level<br />
(TISS) scores prior to transport. Thus, as noted above, confounding <strong>of</strong> differences in patient<br />
535
sickness <strong>and</strong> intensity <strong>of</strong> therapy could account for much <strong>of</strong> the observed variation in transportassociated<br />
morbidity. 24<br />
Opportunity for Impact<br />
A survey conducted in 1990 to review voluntary compliance with the American<br />
Academy <strong>of</strong> Pediatrics (AAP) recommendations to include physicians with higher level <strong>of</strong><br />
training (at least 3 rd year residency) reported that only 28% <strong>of</strong> hospitals with a pediatric critical<br />
care transport team met this recommendation. All teams included a nurse with pediatric<br />
experience <strong>and</strong> a varying degree <strong>of</strong> training, <strong>and</strong> 50% <strong>of</strong> teams included a respiratory therapist. 7<br />
Subchapter 47.1. Interhospital Transport<br />
Study Designs <strong>and</strong> Outcomes<br />
We identified 3 studies with at least a Level 3 study design <strong>and</strong> Level 2 outcomes (see<br />
Table 47.1). Two <strong>of</strong> these studies 10,12 involved pediatric patients. One12 reported the prospective<br />
comparison <strong>of</strong> outcomes for high-risk pediatric patients admitted to two different ICU’s, one <strong>of</strong><br />
which employed a specialized transport team, while the other followed the st<strong>and</strong>ard practice <strong>of</strong><br />
using non-specialized teams. The specialized team consisted <strong>of</strong> a second-year pediatric resident<br />
<strong>and</strong> a pediatric ICU nurse, both trained in pediatric advanced life support, <strong>and</strong> a respiratory<br />
therapist with pediatric experience. Non-specialized teams varied in composition—a physician<br />
was not always present <strong>and</strong> level <strong>of</strong> training in pediatric care for other personnel was not<br />
st<strong>and</strong>ardized. The other pediatric study, from Engl<strong>and</strong>, 10 retrospectively compared outcomes<br />
using a specialized team for the transport <strong>of</strong> intubated newborns from hospitals within 80 miles<br />
to a NICU at a referral center to outcomes during a control period in which transport was<br />
performed by ad hoc doctor/nurse teams. The specialized teams included physicians with more<br />
years <strong>of</strong> experience <strong>and</strong> dedicated transport nurses with specialized training, as well as slight<br />
equipment improvements (humidifier for ventilator <strong>and</strong> invasive/noninvasive blood pressure<br />
monitoring).<br />
The third study (the one involving adults) describes the experience <strong>of</strong> a London teaching<br />
hospital that receives critically ill patients from other facilities by two methods: either<br />
accompanied by the receiving hospital’s special retrieval team consisting <strong>of</strong> an ICU physician,<br />
nurse, <strong>and</strong> medical physics technician (a technician to fix <strong>and</strong> maintain equipment) or st<strong>and</strong>ard<br />
ambulance transport, with an escorting physician supplied by the referring hospital.<br />
The 2 pediatric studies 10,12 reported adverse events during transportation. We counted<br />
adverse events related to intensive care (eg, accidental extubation) as Level 2 <strong>and</strong> physiologic<br />
events (eg, ph < 7.2) as Level 3. (A case could be made for classifying both types <strong>of</strong> adverse<br />
events as Level 3, as neither has a clearly established relationship to adverse events <strong>of</strong> interest).<br />
All studies provided information on case mix in the study <strong>and</strong> control groups.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Although <strong>of</strong> theoretical <strong>and</strong> practical concern, the literature to support the scope,<br />
frequency <strong>and</strong> outcome <strong>of</strong> adverse events during transportation is sparse <strong>and</strong> methodologically<br />
weak. Most studies are small descriptive studies <strong>of</strong> local practices. Factors that limit<br />
comparability between studies include a variety <strong>of</strong> definitions for transport-related adverse<br />
events, unclear descriptions <strong>of</strong> transport team training <strong>and</strong> experience, diverse equipment<br />
536
availability <strong>and</strong> different scoring systems for severity <strong>of</strong> illness (APACHE II, APACHE III,<br />
Glasgow Coma Scale, PRISM, etc). Many confounders affect the evaluation <strong>of</strong> transportation <strong>of</strong><br />
a critically ill patient, among them selection bias, the intervention received at primary hospital,<br />
time spent at primary hospital, adequate stabilization before transport <strong>and</strong> duration <strong>of</strong> transport.<br />
As shown in Table 47.1, 2 studies involving pediatric populations revealed reductions in<br />
intensive care-related adverse events through the use <strong>of</strong> specialized teams for interhospital<br />
transport. 10, 12 In one <strong>of</strong> the studies, patients transported by the st<strong>and</strong>ard (non-specialized) team<br />
were older <strong>and</strong> more likely to have trauma as a diagnosis.12 This difference in patient<br />
populations clearly limits the ability to interpret the results, although the direction <strong>of</strong> bias this<br />
might introduce is not clear. The other pediatric study 10 reported no significant differences in<br />
basic clinical <strong>and</strong> demographic factors between the 2 patient populations, but did not report<br />
PRISM scores.<br />
The single study in adults did not report intensive care-related adverse events, but did<br />
observe significant reductions in surrogate physiologic markers <strong>and</strong> a non-significant reduction<br />
in mortality within 12 hours <strong>of</strong> arrival at the receiving facility. Although an observational study,<br />
there were no differences in the patient populations in terms <strong>of</strong> demographic factors, basic<br />
physiology measurements (FiO2, PaO2, PaCO2, PaO2/FiO2, MAP, heart rate <strong>and</strong> temperature) or<br />
sophisticated measures <strong>of</strong> severity <strong>of</strong> illness (APACHE II, Simplified Acute Physiological<br />
Score-SAPS II).<br />
Studies were underpowered to detect significant mortality differences.<br />
Potential for Harm<br />
A delay in the transfer <strong>of</strong> critically ill patients to referral hospitals because the specialized<br />
team is not available in timely fashion could create a potential for harm although one study<br />
showed no delay or cancellation due to unavailability <strong>of</strong> specialist team. 11<br />
Costs <strong>and</strong> Implementation<br />
Although no firm recommendation can be made, the costs <strong>and</strong> implementation<br />
requirements may only be feasible for tertiary centers that have enough volume to justify the<br />
investment in human <strong>and</strong> physical resources. Time out <strong>of</strong> hospital will vary depending on the<br />
time required to stabilize the patient—not the focus <strong>of</strong> our study. (One study reported an increase<br />
in stabilization time from 80-105 minutes (p
Subchapter 47.2. Intrahospital Transport<br />
Study Designs <strong>and</strong> Outcomes<br />
As shown in Table 47.2, the 3 studies <strong>of</strong> manual versus mechanical ventilation employed<br />
a r<strong>and</strong>omized (or quasi-r<strong>and</strong>omized) controlled design. R<strong>and</strong>omization procedures were not<br />
described in two studies 30, 31 <strong>and</strong> used the last digit <strong>of</strong> the patient record in one study. 32 The<br />
quasi-r<strong>and</strong>omized study 32 implemented a crossover design using manual ventilation or transport<br />
ventilator on one leg <strong>of</strong> the journey <strong>and</strong> vice-versa on the other leg.<br />
Two studies reported on Level 2 <strong>and</strong> 3 outcomes 30, 32 <strong>and</strong> one on level 3 outcomes, 31<br />
venous pressure, oxygen saturation, PetCO2 <strong>and</strong> mean airway pressure during transport. All<br />
studies report on before-after variation, one study 30 also reported on minute variations during the<br />
first 8 minutes <strong>of</strong> transport (see Table 47.2). Only one study reported scores for severity <strong>of</strong><br />
illness (PRISM). 30 Case-mix was inadequately reported in one study 31 <strong>and</strong> not reported in<br />
another. 32<br />
It is worth briefly noting a third practice which involves the use <strong>of</strong> mobile bed/monitor<br />
units versus st<strong>and</strong>ard procedure for intrahospital transportation. Studies on this topic are limited<br />
to descriptive experiences <strong>of</strong> local practice with no definition or systematic evaluation <strong>of</strong> adverse<br />
events, 17,33-35 so these practices were not reviewed further.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
The clinical significance <strong>of</strong> the hyperventilation observed in manually ventilated patients<br />
during intrahospital transportation has yet to be determined. Mechanical ventilation was<br />
associated with respiratory alkalosis when precision <strong>of</strong> ventilatory settings was inaccurate. No<br />
adverse effect, (ie, morbidity) was observed as a result <strong>of</strong> the method <strong>of</strong> ventilation. Use <strong>of</strong> a<br />
volumeter when manually ventilating a patient reduced the risk <strong>of</strong> hyperventilation. Studies were<br />
underpowered to detect significant mortality differences.<br />
538
Potential for Harm<br />
Inadequate maintenance <strong>and</strong>/or precision <strong>of</strong> transport ventilator may create an opportunity for<br />
harm.<br />
Costs <strong>and</strong> Implementation<br />
Portable mechanical ventilators are much more expensive, require more hours <strong>of</strong> training to<br />
manipulate <strong>and</strong> frequent use to maintain experience. Costs were not mentioned.<br />
Comment<br />
One r<strong>and</strong>omized controlled trial in pediatric postoperative cardiac patients showed an increase in<br />
markers <strong>of</strong> hyperventilation for patients in the manually-ventilated group. Otherwise, manual<br />
ventilation appears to achieve results comparable to portable mechanical ventilation. Use <strong>of</strong> a<br />
volumeter when manually ventilating patients, <strong>and</strong> the addition <strong>of</strong> a blender to reduce FiO2 when<br />
manually ventilating neonates, 30 may adequately reduce the risks <strong>of</strong> hyperventilation, rendering<br />
mechanical ventilation during intrahospital transport unnecessary.<br />
539
Table 47.1. Specialized transport teams versus st<strong>and</strong>ard care for interhospital transport<br />
Study Setting Study Design,<br />
Outcomes<br />
<strong>Critical</strong>ly ill children transported to two<br />
PICU’s in Albany, NY (specialized<br />
transport team) <strong>and</strong> Syracuse, NY<br />
(st<strong>and</strong>ard care): 1992-94 12<br />
Intubated newborns transported by<br />
specialized doctor/nurse team (increased<br />
training <strong>and</strong> experience) to NICU in<br />
Nottingham, Engl<strong>and</strong>: 1994-95, <strong>and</strong><br />
historical control period during which<br />
non-specialized (ad hoc) doctor/nurse<br />
team transported patients to the same<br />
NICU: 1991-93 10<br />
<strong>Critical</strong>ly ill adults transported to a<br />
university ICU in London: 1996-1997;<br />
specialist team with mobile ICU<br />
compared with emergency ambulance<br />
with medical escort 11<br />
Level 3,<br />
Level 2<br />
Level 3,<br />
Levels 2&3†<br />
Level 3,<br />
Levels 1&3<br />
540<br />
Main Results<br />
Significant decrease adverse event<br />
related to intensive care for patients<br />
transported with specialized team<br />
compared to st<strong>and</strong>ard care: 1/47 (2%)<br />
vs. 18/92 (20%), p0.05<br />
Nonsignificant reduction in<br />
endotracheal tube-related events<br />
(blocked or dislodged endotracheal<br />
tubes): 0/146 (95% CI: 0-3.2%)<br />
patients transported by specialized<br />
teams vs. 3/73 (4.1%, 95% CI: 1.1-<br />
12.3%) by ad hoc teams<br />
Reductions also observed in adverse<br />
physiologic end-points; such as<br />
abnormal ph (p
Table 47.2 Manual versus mechanical ventilation for intrahospital transportation<br />
Study Setting Study<br />
Design,<br />
Outcomes<br />
30 ventilator dependent,<br />
critically ill adults in ICU:<br />
manually ventilated with<br />
self-inflating bag group,<br />
manually ventilated selfinflating<br />
bag with volumeter,<br />
<strong>and</strong> mechanically ventilated<br />
group (Federal Republic <strong>of</strong><br />
Germany) 31<br />
28 critically ill adults <strong>and</strong><br />
adolescents transported from<br />
emergency department for<br />
diagnostic procedures in a<br />
University Hospital: manual<br />
ventilation versus transport<br />
ventilator (US) 32<br />
51 pediatric postoperative<br />
cardiac surgery patients who<br />
were transported within<br />
hospital while intubated:<br />
manually ventilated versus<br />
mechanically ventilated<br />
(US) 30<br />
Level 1,<br />
Level 3<br />
Level 1,<br />
Level 2&3<br />
Level 1,<br />
Level 2&3<br />
Main Results<br />
Pre/post transport: PaCO2 decreased from 41±2 to<br />
34±2 (p
References<br />
1. Pollack MM, Alex<strong>and</strong>er SR, Clarke N, Ruttimann UE, Tesselaar HM, Bachulis AC.<br />
Improved outcomes from tertiary center pediatric intensive care: a statewide comparison <strong>of</strong><br />
tertiary <strong>and</strong> nontertiary care facilities. Crit Care Med. 1991;19:150-159.<br />
2. Kanter RK, Tompkins JM. Adverse events during interhospital transport: physiologic<br />
deterioration associated with pretransport severity <strong>of</strong> illness. Pediatrics. 1989;84:43-48.<br />
3. Borker S, Rudolph C, Tsuruki T, Williams M. Interhospital referral <strong>of</strong> high-risk newborns<br />
in a rural regional perinatal program. J Perinatol. 1990;10:156-163.<br />
4. McCloskey KA, Faries G, King WD, Orr RA, Plouff RT. Variables predicting the need for<br />
a pediatric critical care transport team. Pediatr Emerg Care. 1992;8:1-3.<br />
5. Young JS, Bassam D, Cephas GA, Brady WJ, Butler K, Pomphrey M. Interhospital versus<br />
direct scene transfer <strong>of</strong> major trauma patients in a rural trauma system. Am Surg.<br />
1998;64:88-91; discussion 91-92.<br />
6. Dryden CM, Morton NS. A survey <strong>of</strong> interhospital transport <strong>of</strong> the critically ill child in the<br />
United Kingdom. Paediatr Anaesth. 1995;5:157-160.<br />
7. McCloskey KA, Johnston C. Pediatric critical care transport survey: team composition <strong>and</strong><br />
training, mobilization time, <strong>and</strong> mode <strong>of</strong> transportation. Pediatr Emerg Care. 1990;6:1-3.<br />
8. Gentleman D, Jennett B. Audit <strong>of</strong> transfer <strong>of</strong> unconscious head-injured patients to a<br />
neurosurgical unit. Lancet. 1990;335:330-334.<br />
9. Domeier RM, Hill JD, Simpson RD. The development <strong>and</strong> evaluation <strong>of</strong> a paramedicstaffed<br />
mobile intensive care unit for inter-facility patient transport. Prehospital Disaster<br />
Med. 1996;11:37-43.<br />
10. Leslie AJ, Stephenson TJ. Audit <strong>of</strong> neonatal intensive care transport–closing the loop. Acta<br />
Paediatr. 1997;86:1253-1256.<br />
11. Bellingan G, Olivier T, Batson S, Webb A. Comparison <strong>of</strong> a specialist retrieval team with<br />
current United Kingdom practice for the transport <strong>of</strong> critically ill patients. Intensive Care<br />
Med. 2000;26:740-744.<br />
12. Edge WE, Kanter RK, Weigle CG, Walsh RF. Reduction <strong>of</strong> morbidity in interhospital<br />
transport by specialized pediatric staff. Crit Care Med. 1994;22:1186-1191.<br />
13. Hurst JM, Davis K, Johnson DJ, Branson RD, Campbell RS, Branson PS. Cost <strong>and</strong><br />
complications during in-hospital transport <strong>of</strong> critically ill patients: a prospective cohort<br />
study. J Trauma. 1992;33:582-585.<br />
14. Szem JW, Hydo LJ, Fischer E, Kapur S, Klemperer J, Barie PS. High-risk intrahospital<br />
transport <strong>of</strong> critically ill patients: safety <strong>and</strong> outcome <strong>of</strong> the necessary "road trip". Crit<br />
Care Med. 1995;23:1660-1666.<br />
15. Wallen E, Venkataraman ST, Grosso MJ, Kiene K, Orr RA. Intrahospital transport <strong>of</strong><br />
critically ill pediatric patients. Crit Care Med. 1995;23:1588-1595.<br />
16. Indeck M, Peterson S, Smith J, Brotman S. Risk, cost, <strong>and</strong> benefit <strong>of</strong> transporting ICU<br />
patients for special studies. J Trauma. 1988;28:1020-1025.<br />
17. Taylor JO, Chulay, L<strong>and</strong>ers CF, Hood W, Abelman WH. Monitoring high-risk cardiac<br />
patients during transportation in hospital. Lancet. 1970;2:1205-1208.<br />
18. Macnab AJ. Optimal escort for interhospital transport <strong>of</strong> pediatric emergencies. J Trauma.<br />
1991;31:205-209.<br />
19. Barry PW, Ralston C. Adverse events occurring during interhospital transfer <strong>of</strong> the<br />
critically ill. Arch Dis Child. 1994;71:8-11.<br />
542
20. Duke G, Green J. Outcome <strong>of</strong> critically ill patients undergoing interhospital transfer.<br />
<strong>Medical</strong> Journal <strong>of</strong> Australia. 2001;174:122-125.<br />
21. Smith I, Fleming S, Cernaianu A. Mishaps during transport from the intensive care unit.<br />
Crit Care Med. 1990;18:278-281.<br />
22. Andrews PJ, Piper IR, Dearden NM, Miller JD. Secondary insults during intrahospital<br />
transport <strong>of</strong> head-injured patients. Lancet. 1990;335:327-330.<br />
23. Braman SS, Dunn SM, Amico CA, Millman RP. Complications <strong>of</strong> intrahospital transport in<br />
critically ill patients. Ann Intern Med. 1987;107:469-473.<br />
24. Kanter RK, Boeing NM, Hannan WP, Kanter DL. Excess morbidity associated with<br />
interhospital transport. Pediatrics. 1992;90:893-898.<br />
25. Kalisch BJ, Kalisch PA, Burns SM, Kocan MJ, Prendergast V. Intrahospital transport <strong>of</strong><br />
neuro ICU patients. J Neurosci Nurs. 1995;27:69-77.<br />
26. Insel J, Weissman C, Kemper M, Askanazi J, Hyman AI. Cardiovascular changes during<br />
transport <strong>of</strong> critically ill <strong>and</strong> postoperative patients. Crit Care Med. 1986;14:539-542.<br />
27. Bion JF, Wilson IH, Taylor PA. Transporting critically ill patients by ambulance: audit by<br />
sickness scoring. Br Med J (Clin Res Ed). 1988;296:170.<br />
28. Waddell G, Scott PD, Lees NW, Ledingham IM. Effects <strong>of</strong> ambulance transport in<br />
critically ill patients. Br Med J. 1975;1:386-389.<br />
29. Britto J, Nadel S, Maconochie I, Levin M, Habibi P. Morbidity <strong>and</strong> severity <strong>of</strong> illness<br />
during interhospital transfer: impact <strong>of</strong> a specialised paediatric retrieval team. BMJ.<br />
1995;311:836-839.<br />
30. Dockery WK, Futterman C, Keller SR, Sheridan MJ, Akl BF. A comparison <strong>of</strong> manual <strong>and</strong><br />
mechanical ventilation during pediatric transport. Crit Care Med. 1999;27:802-806.<br />
31. Gervais HW, Eberle B, Konietzke D, Hennes HJ, Dick W. Comparison <strong>of</strong> blood gases <strong>of</strong><br />
ventilated patients during transport. Crit Care Med. 1987;15:761-763.<br />
32. Hurst JM, Davis K, Branson RD, Johannigman JA. Comparison <strong>of</strong> blood gases during<br />
transport using two methods <strong>of</strong> ventilatory support. J Trauma. 1989;29:1637-1640.<br />
33. Link J, Krause H, Wagner W, Papadopoulos G. Intrahospital transport <strong>of</strong> critically ill<br />
patients. Crit Care Med. 1990;18:1427-1429.<br />
34. Holst D, Rudolph P, Wendt M. Mobile workstation for anaesthesia <strong>and</strong> intensive-care<br />
medicine. Lancet. 2000;355:1431-1432.<br />
35. Hanning CD, Gilmour DG, Hothersal AP, Aitkenhead AR, Venner RM, Ledingham IM.<br />
Movement <strong>of</strong> the critically ill within hospital. Intensive Care Med. 1978;4:137-143.<br />
543
Section H. Role <strong>of</strong> the <strong>Patient</strong><br />
Chapter 48. Procedures For Obtaining Informed Consent<br />
Chapter 49. Advance Planning For End-<strong>of</strong>-Life Care<br />
Chapter 50. Other <strong>Practices</strong> Related to <strong>Patient</strong> Participation<br />
544
545
Chapter 48. Procedures For Obtaining Informed Consent<br />
Laura T. Pizzi, PharmD<br />
Neil I. Goldfarb<br />
David B. Nash, MD, MBA<br />
Thomas Jefferson University School <strong>of</strong> Medicine<br />
<strong>and</strong> Office <strong>of</strong> Health Policy & Clinical Outcomes<br />
Background<br />
The process <strong>of</strong> obtaining informed consent, whether a written document or an oral<br />
communication is one means <strong>of</strong> ensuring that patients underst<strong>and</strong> the risks <strong>and</strong> benefits <strong>of</strong> a<br />
treatment or medical intervention. Rooted in medical ethics <strong>and</strong> codified as a legal principle, it is<br />
based on the assertion that a competent individual has the right to determine what will or will not<br />
be done to him or her. 1 The American <strong>Medical</strong> Association (AMA) Code <strong>of</strong> <strong>Medical</strong> Ethics<br />
establishes informed consent as an ethical obligation <strong>of</strong> physicians. 2 In addition to being an<br />
ethical obligation <strong>of</strong> physicians, legislation in all 50 states requires that patients be informed <strong>of</strong><br />
all important aspects <strong>of</strong> a treatment <strong>and</strong>/or procedures, although the details <strong>of</strong> these laws <strong>and</strong><br />
statutes differ greatly. 2 Failure to obtain adequate informed consent renders a physician liable for<br />
negligence or battery3 <strong>and</strong> constitutes medical malpractice.<br />
To date, studies <strong>of</strong> informed consent have not investigated outcomes related to the<br />
adequacy <strong>of</strong> the communication, ins<strong>of</strong>ar as this may impact patient safety. 4,5 Physician-patient<br />
communication styles have been linked to lower rates <strong>of</strong> malpractice claims. 6 Nonetheless, as<br />
noted by Levinson, malpractice claims do not reflect the actual rate <strong>of</strong> negligence. While some<br />
have hypothesized that better informed consent could improve the patient-physician relationship,<br />
establish trust, increase patient compliance, <strong>and</strong> provide information that could reduce medical<br />
error, this has not been shown. 7,8 In the absence <strong>of</strong> a direct link between adequate informed<br />
consent <strong>and</strong> the reduction <strong>of</strong> medical error, the “patient safety outcome” reviewed in this chapter<br />
is the patient’s provision <strong>of</strong> adequate informed consent.<br />
Practice Description<br />
Informed consent is a process through which a physician informs a patient about the risks<br />
<strong>and</strong> benefits <strong>of</strong> a proposed therapy <strong>and</strong> allows the patient to decide whether the therapy will be<br />
undertaken. 9 It may be received in one sitting, or over a period <strong>of</strong> time, 7 either orally or in<br />
writing or a combination <strong>of</strong> the two. Informed consent procedures have been instituted in both<br />
research <strong>and</strong> clinical medicine. In the former case, Federal regulations establish strict guidelines<br />
for informed consent that are monitored by a special board at each institution (Institutional<br />
Review Board). In addition, risks <strong>and</strong> adverse events that occur while research is in progress are<br />
followed closely <strong>and</strong> reported. As such informed consent in the research setting differs greatly<br />
from informed consent in the clinical setting. 10<br />
In clinical practice, formal efforts, such as the signing <strong>of</strong> a consent form, (presumably<br />
preceded by adequate exchange <strong>of</strong> information), are only undertaken in some circumstances,<br />
notably prior to major invasive procedures such as radiologic procedures <strong>and</strong> surgery. Less well<br />
appreciated is that all medical care, including pharmacy prescriptions or laboratory tests, requires<br />
informal informed consent, except when the patient is incompetent to make a decision or<br />
relinquishes the right to provide it. 3 Studies suggest that in practice only minimal formal efforts<br />
are made to obtain informed consent for routine interventions. 11, 12<br />
546
Legislation governing the requirements <strong>of</strong>, <strong>and</strong> conditions under which, consent must be<br />
obtained varies greatly from State to State. General guidelines, such as those proposed by the<br />
AMA require patients to be informed <strong>of</strong> the nature <strong>of</strong> their condition <strong>and</strong> the proposed<br />
procedure, the purpose <strong>of</strong> the procedure, the risks <strong>and</strong> benefits <strong>of</strong> the proposed treatments, the<br />
probability <strong>of</strong> the anticipated risks <strong>and</strong> benefits, alternatives to the treatment <strong>and</strong> the associated<br />
risks <strong>and</strong> benefits, <strong>and</strong> the risks <strong>and</strong> benefits <strong>of</strong> not receiving the treatment or procedure. 2,3,13<br />
As discussed below, procedures to obtain informed consent may not adequately promote<br />
the patient’s comprehension <strong>of</strong> the information provided, rendering the consent not truly<br />
“informed.” Interventions that may prove beneficial in improving <strong>and</strong> ensuring the patient‘s<br />
underst<strong>and</strong>ing include redrafting <strong>of</strong> consent forms to reduce complexity, providing written<br />
materials to accompany oral conversations, using multimedia or other techniques to improve<br />
comprehension, <strong>and</strong> asking patients to recap discussions about the procedure.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> Target <strong>Safety</strong> Problem<br />
Procedures to obtain consent must ensure that the patient underst<strong>and</strong>s his or her<br />
condition, as well as the risk <strong>and</strong> benefits <strong>of</strong> treatment, <strong>and</strong> its alternatives. It has been estimated<br />
that less than half <strong>of</strong> the US population underst<strong>and</strong>s commonly used medical terms. 14,15 This<br />
“health literacy” problem may impact the ability <strong>of</strong> the patient to underst<strong>and</strong> any attempts to<br />
obtain information. In addition to lack <strong>of</strong> comprehension, procedures to obtain informed consent<br />
may be incomplete.<br />
Several studies have noted the various insufficiencies in procedures to obtain informed<br />
consent. Three studies examined the completeness <strong>of</strong> physician-patient conversations in<br />
obtaining informed consent. Braddock et al 12 focused on outpatient discussions. Recognizing<br />
that some procedures may require more discussion than others, they created a three-tiered<br />
evaluation procedure, in which the completeness <strong>of</strong> patient-physician discussion differ according<br />
to the complexity <strong>of</strong> the decision being discussed. Basic decisions, such as laboratory tests,<br />
require the least in-depth discussion, covering only the patient’s role, the clinical nature <strong>of</strong> the<br />
decision, <strong>and</strong> exploration <strong>of</strong> patient preferences. Intermediate decisions, such as changes in<br />
medication, require a moderate depth <strong>of</strong> discussion, incorporating the Tier 1 subjects, <strong>and</strong> adding<br />
a discussion <strong>of</strong> alternative treatments, the risks <strong>and</strong> benefits <strong>of</strong> the alternatives, <strong>and</strong> an<br />
assessment <strong>of</strong> the patients underst<strong>and</strong>ing. Complex decisions such as surgery require a<br />
discussion <strong>of</strong> the uncertainties associated with the decision, in addition to all <strong>of</strong> the<br />
aforementioned steps. Analyzing audiotaped conversations between 1057 patients <strong>and</strong> 59<br />
primary-care physicians <strong>and</strong> 65 surgeons, Braddock et al found that 17.2% <strong>of</strong> basic decisions<br />
contained all required components, while none <strong>of</strong> the intermediate decisions <strong>and</strong> only one <strong>of</strong> the<br />
complex decisions contained all <strong>of</strong> the required components. Applying only the Tier 1 basic<br />
consent st<strong>and</strong>ards, 20.5% <strong>of</strong> basic decisions, 21.9% <strong>of</strong> intermediate decisions, <strong>and</strong> 38.2% <strong>of</strong><br />
complex decisions met all the criteria.<br />
In a study <strong>of</strong> informed consents <strong>of</strong> surrogates for pediatric patients undergoing surgery,<br />
coders examined audiotaped conversations with surrogates, as well as structured interview <strong>and</strong><br />
questionnaire data regarding the conversations. They noted that patients’ recall <strong>of</strong> the<br />
conversations with physicians <strong>of</strong>ten omitted key components, such as the risks <strong>and</strong> benefits <strong>of</strong><br />
the procedures. 16<br />
Bottrell et al 13 examined the completeness <strong>of</strong> 540 consent forms from 157 hospitals<br />
nationwide. Of these, 26.4% included all four <strong>of</strong> the basic elements (risks, benefits, alternatives,<br />
<strong>and</strong> other important aspects <strong>of</strong> the procedure). Eighty-seven percent noted the general possibility<br />
<strong>of</strong> risk, but less than half provided specific information. Alternatives were noted in 56.9% <strong>of</strong> the<br />
547
forms, <strong>and</strong> benefits appeared in 37%, though most <strong>of</strong> these were general references rather than<br />
specific information. Although 74% <strong>of</strong> consent forms were deemed incomplete, it is unknown<br />
whether physician-patient discussions that preceded the signing <strong>of</strong> the consent form included the<br />
missing information.<br />
A study by Mark et al9 found that 82.4% <strong>of</strong> 102 participants reported that they<br />
understood everything that their physicians had described about a procedure <strong>and</strong> indicated that<br />
all <strong>of</strong> their questions had been answered. Eighteen patients had remaining unanswered questions.<br />
Half <strong>of</strong> this group requested more time to speak with their physicians, while the other 9 felt that<br />
their questions were not important.<br />
In a study by Lavelle-Jones, 17 69% <strong>of</strong> patients admitted that they did not read a consent<br />
form before signing it. In addition, approximately half <strong>of</strong> the patients awaiting treatment were<br />
unhappy with the amount <strong>of</strong> information they received, with 21% stating that most <strong>of</strong> the<br />
information they obtained about their surgical treatment was obtained outside <strong>of</strong> the hospital.<br />
Consent forms have been targeted for their lack <strong>of</strong> readability. <strong>Patient</strong>s with limited<br />
reading ability are at increased risk for medical errors, due to problems reading medication<br />
bottles, appointment slips, self-care instructions, <strong>and</strong> health education brochures. 18 These<br />
patients may also have trouble reading materials intended to aid in obtaining informed consent.<br />
According to the National Adult Literacy Survey <strong>of</strong> 1993, approximately 40-44 million<br />
Americans were functionally illiterate, defined as the inability to complete basic reading tasks<br />
required to function as a member <strong>of</strong> society. 19 In addition, even in educated adults, the highest<br />
grade-level completed may not reflect actual reading comprehension level. A study <strong>of</strong> 100 adult<br />
cancer patients found that most read at a mean grade-level equivalent <strong>of</strong> between 10 th <strong>and</strong> 11 th<br />
grade. The authors suggest that forms <strong>and</strong> educational materials be written at a grade-level three<br />
levels below the highest level <strong>of</strong> education completed by a patient. 20<br />
Several studies have examined the readability <strong>of</strong> procedure consent forms. Two studies<br />
examined consent forms for radiologic procedures using computer generated “readability”<br />
scores. Consent forms for use with iodinated contrast media found that 12.35 years <strong>of</strong> education<br />
were required to read consent forms. 21 A similar study <strong>of</strong> general radiologic procedure consent<br />
forms found that they required a mean <strong>of</strong> 15 years <strong>of</strong> education. Only 16% <strong>of</strong> forms could be<br />
understood by patients with a high-school education. 11 Another Hopper et al study22 found that<br />
general hospital consent forms were written at a grade level <strong>of</strong> 12.6. Just over half could be<br />
understood by a patient with a high school education, less than a third by patients with a 10 th<br />
grade reading level, <strong>and</strong> just over 5% by patients with an 8 th grade reading level.<br />
Opportunities for Impact<br />
While informed consent is a well-established practice, it <strong>of</strong>ten fails to meet its stated<br />
purpose. Several methods <strong>of</strong> improving the procedures <strong>of</strong> obtaining informed consent have been<br />
proposed, including improving the readability <strong>of</strong> consent forms, 22 asking patients for recall to<br />
establish underst<strong>and</strong>ing, 3,23 adding additional stimuli, such as multimedia presentations 24 <strong>and</strong><br />
providing written information. 17<br />
Lavelle-Jones 17 found that elderly patients (over 60 years <strong>of</strong> age) had poorer recall than<br />
younger patients. In addition, patients with internal locus <strong>of</strong> control—those who believed their<br />
health was in their own control—were better informed than those with an external locus <strong>of</strong><br />
control. <strong>Patient</strong>s with above average IQ exhibited better recall. These findings could indicate “atrisk”<br />
groups that interventions may target.<br />
One author argues that an important opportunity for impact is to change the model <strong>of</strong><br />
implementing informed consent from a single event approach to a process approach. 7 Currently<br />
548
obtaining informed consent <strong>of</strong>ten revolves around the signing <strong>of</strong> the consent form. Although this<br />
approach clearly delineates the responsibility <strong>of</strong> health care providers, provides documentation<br />
<strong>of</strong> the consent, <strong>and</strong> fits easily into the current provider structure, it <strong>of</strong>ten results in patients failing<br />
to actually comprehend the information <strong>and</strong> reinforces physicians’ conception that the consent<br />
ritual is futile. In contrast, a process model involves the physician providing information over<br />
time, establishing a better patient-physician relationship <strong>and</strong> better comprehension <strong>of</strong> medical<br />
care. As <strong>of</strong> yet there is no data to support these suppositions.<br />
Since the definition <strong>of</strong> adequate informed consent is debatable, the number <strong>of</strong> individuals<br />
currently not receiving interventions to obtain adequate informed consent is likely to be quite<br />
high, but is not known.<br />
Study Design, Outcomes <strong>and</strong> Effectiveness <strong>of</strong> Various Approaches<br />
Improving Readability <strong>of</strong> Consent Forms <strong>and</strong> Education Materials<br />
In order to ensure that patients underst<strong>and</strong> the procedure to which they are consenting, it<br />
is important that all materials be presented in a comprehensible manner. Consent forms are<br />
written with relatively complex sentence structure <strong>and</strong> vocabulary, making it difficult for the<br />
average adult to interpret the information. In addition, providing consent forms in the primary<br />
language <strong>of</strong> patients may improve consent procedures. However, we located no studies<br />
examining the effectiveness <strong>of</strong> such practices.<br />
Structured Discussions<br />
Informed consent is <strong>of</strong>ten obtained during informal discussion between physicians or<br />
nurses <strong>and</strong> patients, <strong>and</strong> (as discussed above), these discussions frequently do not cover all <strong>of</strong> the<br />
relevant information. 12 Two studies examined the use <strong>of</strong> a structured interview format in<br />
providing information to patients. Solomon et al 25 studied 36 patients receiving cardiac<br />
catheterization for the first time at a Veterans hospital. <strong>Patient</strong>s were r<strong>and</strong>omized into two groups<br />
(Level 1 design). Both groups were briefed by a cardiologist regarding the procedure (st<strong>and</strong>ard<br />
care). In addition, the experimental group received a 30-minute structured teaching session with<br />
a nurse to discuss all aspects <strong>of</strong> the procedure, including the purposes, techniques, sensations,<br />
risks <strong>and</strong> benefits. <strong>Patient</strong>s in the experimental group also received an illustrated guide to<br />
cardiovascular procedures <strong>and</strong> an educational pamphlet. All subjects were tested using a 13-item<br />
questionnaire covering the information that should have been imparted during informed consent<br />
procedures (Level 3 outcome). The intervention group scored significantly better than the control<br />
group (11.5 vs. 8.9).<br />
In a later study, Dawes et al 26 studied 190 hospitalized patients undergoing ENT surgery.<br />
<strong>Patient</strong>s were assigned to one <strong>of</strong> 4 groups. Groups 1 <strong>and</strong> 2 were assigned at different times<br />
(Level 2 design), while groups 3 <strong>and</strong> 4 were r<strong>and</strong>omly assigned (Level 1 design). Group 1 had<br />
no consent interview until after assessment at the conclusion <strong>of</strong> the study. Group 2 engaged in an<br />
informal interview with a physician (reflecting current practice), during which the number <strong>of</strong><br />
complications discussed was recorded. Group 3 engaged in a structured interview with a<br />
physician, covering the purpose <strong>and</strong> technique <strong>of</strong> the procedure, complications, sensations,<br />
alternatives, <strong>and</strong> benefits (a checklist was used as a guide). The final group, Group 4, engaged in<br />
the same structured interview, except that the patient was provided with a copy <strong>of</strong> the checklist<br />
<strong>and</strong> allowed to take it with them after the interview. <strong>Patient</strong> anxiety was assessed following<br />
consent using a visual analog scale, then reassessed at an interview, given 4 hours later. At the<br />
later interview, patients were asked to recount orally the operation name, describe what would be<br />
549
done, list complications, <strong>and</strong> state whether they understood the information given to them (Level<br />
3 outcome). All but Group 1 (control group) patients showed a drop to normal anxiety after<br />
informed consent. Group 2 (informal discussion) remembered proportionately more<br />
complications mentioned in the informal interview, although fewer complication were covered.<br />
Groups 3 <strong>and</strong> 4 recalled more total complications than Group 2. There was no differences<br />
between Groups 3 <strong>and</strong> 4 (structured interviews).<br />
Asking for Recall<br />
A simple method <strong>of</strong> determining whether a patient underst<strong>and</strong>s information regarding a<br />
procedure is to ask the patient to recount what he or she has been told. 3 Two studies examined<br />
this intervention, using a r<strong>and</strong>omized controlled trial design (Level 1 design). In the first study,<br />
informed consent was solicited from 50 patients undergoing percutaneous lung biopsy. 23<br />
Twenty-seven patients in the control group were <strong>of</strong>fered the st<strong>and</strong>ard procedure for obtaining<br />
informed consent: approximately 30 minutes prior to the procedure, the physician described the<br />
procedure in detail, including its risks <strong>and</strong> benefits. Four complications were specifically<br />
described, along with their relative risks. <strong>Patient</strong>s were asked to sign a st<strong>and</strong>ard consent form.<br />
Twenty-three patients received the same procedure, but in addition, they were asked to describe<br />
all 4 potential complications. The procedure was repeated until all patients in the intervention<br />
group could recount all <strong>of</strong> the complications. This modified procedure usually took less than 5<br />
minutes extra compared with the traditional approach. <strong>Patient</strong>s were interviewed 2 hours after<br />
the procedure was completed. <strong>Patient</strong>s in the modified consent group had better recall (Level 2<br />
outcome) than patients in the control group (56% vs. 14% with high recall (recalling 3-4 out <strong>of</strong> 4<br />
risks), <strong>and</strong> 13% vs. 44% with low recall (recalling 0-1 out <strong>of</strong> 4 risks)).<br />
A second study examined verbalization in 20 patients undergoing anterior cruciate<br />
ligament (ACL) reconstruction. 27 <strong>Patient</strong>s were r<strong>and</strong>omly assigned to 2 groups. Both groups<br />
received the st<strong>and</strong>ard education for ACL reconstruction, which included the use <strong>of</strong> a 3-D model<br />
<strong>of</strong> the knee, discussion with a physician about the procedure, <strong>and</strong> obtaining informed consent.<br />
The experimental group (8 subjects) also were asked to repeat back the risks <strong>of</strong> the procedure<br />
until they could accurately recall all risks discussed. One-month later, recall regarding the<br />
information received was tested using a 3-item questionnaire (Level 2 outcome). All 8 in the<br />
experimental group answered all questions correctly, while only four out <strong>of</strong> the 12 in the control<br />
group answered all questions correctly (p=0.03).<br />
Use <strong>of</strong> Visual or Auditory Learning Aids<br />
Adding additional stimuli may increase the ability <strong>of</strong> the patient to underst<strong>and</strong><br />
information being conveyed or increase retention <strong>of</strong> that information. For instance, use <strong>of</strong> visual<br />
diagrams may make a procedure easier to underst<strong>and</strong>. Multi-media may also promote better<br />
comprehension. Three studies have examined the addition <strong>of</strong> visual stimuli for informed consent.<br />
One study <strong>of</strong> patients undergoing back surgery examined the impact <strong>of</strong> a diagnosis-specific<br />
videodisk program on patient outcomes. 24 <strong>Patient</strong>s (n=393) who were c<strong>and</strong>idates for elective<br />
back surgery (primarily for herniated disc <strong>and</strong> spinal stenosis) were r<strong>and</strong>omized to two education<br />
groups (Level 1 design). The control group received a written booklet regarding the surgical <strong>and</strong><br />
non-surgical treatments for herniated disc <strong>and</strong> spinal stenosis, a description <strong>of</strong> expected<br />
outcomes, <strong>and</strong> a short self-test on the booklet information. <strong>Patient</strong>s in the experimental group<br />
also received the booklet, but in addition viewed a videodisk program. The program allowed a<br />
patient to enter their diagnosis <strong>and</strong> age <strong>and</strong> receive customized information on the alternative<br />
treatments, discussion <strong>of</strong> the diagnoses (<strong>and</strong> the ambiguities <strong>of</strong> diagnosis), <strong>and</strong> interviews from<br />
550
patients that had been treated surgically <strong>and</strong> non-surgically. At the end the patient was provided<br />
with a printout <strong>of</strong> outcome probabilities. <strong>Patient</strong>s in the experimental group were slightly more<br />
likely to report that they had all the information they wanted (Level 3 outcome). Rates <strong>of</strong><br />
patients consenting to surgery for herniated disk (32% vs. 47%) <strong>and</strong> other diagnoses (5.4% vs.<br />
14.0%) were lower in the videodisk group.<br />
Hopper et al28 also tested an interactive video program, although the tested program was<br />
computer based. One-hundred <strong>and</strong> sixty outpatients referred for IV contrast media studies were<br />
stratified by age, sex, <strong>and</strong> previous exposure to contrast media, then r<strong>and</strong>omized to receive either<br />
a written consent form, or an interactive computer-based video (Level 1 design). Subjects in the<br />
control group received a consent form designed to be read at an eighth grade reading level.<br />
Subjects in the experimental group viewed a video in which a physician used identical words as<br />
the consent form to inform subjects about the procedure <strong>and</strong> risks. Subjects then had an option <strong>of</strong><br />
hearing more about the risks. If subjects chose not to hear additional information, they were<br />
provided with printouts <strong>of</strong> the risks (with the minimal information already given). Otherwise<br />
subjects were provided with printouts certifying that they had completed the program. All<br />
subjects were tested using a 7-item questionnaire regarding the procedure <strong>and</strong> risks (Level 2<br />
outcome). <strong>Patient</strong>s that viewed the video responded correctly more <strong>of</strong>ten than the control group<br />
when asked about general aspects <strong>of</strong> the procedure. Female patients in the video group also<br />
responded correctly more <strong>of</strong>ten to questions about risks, although this finding did not hold for<br />
male patients. The video did take approximately 1.6 minutes <strong>of</strong> additional time to complete, <strong>and</strong><br />
there was no difference in patients desire for additional knowledge between the groups (Level 3<br />
outcome).<br />
One final study did not use an interactive video, but tested whether providing information<br />
via video was superior to providing information via an informal discussion with the physician<br />
(st<strong>and</strong>ard practice). 29 Two-hundred <strong>and</strong> twenty four subjects referred for colonoscopy were<br />
stratified into previous or no previous colonoscopy, then r<strong>and</strong>omized into 3 groups (Level 1<br />
design). The control group had a structured discussion with a physician, in which the physician<br />
covered the same information covered in the video according to a checklist. The video-only<br />
group watched a 5-minute videotape, in which a physician described the procedure, <strong>and</strong> its risk<br />
<strong>and</strong> benefits. The third group (video <strong>and</strong> discussion), watched the video <strong>and</strong> then engaged in a<br />
structured discussion with a physician. All patients were tested using a 13-item questionnaire<br />
(Level 2 outcome). Both video groups gave more correct responses than the discussion-only<br />
group, although they did not differ from each other. In addition, patient anxiety levels were<br />
measured using the State-Trait Anxiety Inventory (Level 3 outcome). There were no differences<br />
among the 3 groups.<br />
Providing Written information to <strong>Patient</strong>s<br />
Providing written information to patients regarding their diagnoses, proposed treatments,<br />
<strong>and</strong> other information given during informed consent discussion allows the patient to refer back<br />
to such information, <strong>and</strong> possibly increases comprehension. One early study compared the thencommon<br />
practice, informal interview between patient <strong>and</strong> doctor, with provision <strong>of</strong> a written<br />
consent form. 30 Eighty patients, referred for a first excretory urography, were r<strong>and</strong>omly assigned<br />
to 2 groups (Level 1 design). The control group received st<strong>and</strong>ard care, the informal interview.<br />
The experimental group received the same interview along with a detailed written consent form<br />
to read <strong>and</strong> sign. <strong>Patient</strong> clinical reactions <strong>and</strong> side effects were monitored (Level 1 outcome).<br />
One to 3 days after the procedures, patients retrospectively rated discomfort, fear or<br />
apprehension, <strong>and</strong> underst<strong>and</strong>ing <strong>of</strong> the procedure (Level 3 outcome). <strong>Patient</strong>s also completed an<br />
551
8-item knowledge examination covering the information on the consent form (Level 2 outcome),<br />
or in the informal interview. Experimental subjects scored significantly better (p=
References<br />
1. Edwards SJ, Lilford RJ, Thornton J, et al. Informed consent for clinical trials: in search <strong>of</strong><br />
the "best" method. Soc Sci Med. 1998;47:1825-40.<br />
2. American <strong>Medical</strong> Association. Informed consent. American <strong>Medical</strong> Association. April<br />
11, 2001.<br />
3. Pape T. Legal <strong>and</strong> ethical considerations <strong>of</strong> informed consent. AORN J. 1997;65:1122-7.<br />
4. Stewart MA. Effective physician-patient communication <strong>and</strong> health outcomes: a review.<br />
CMAJ. 1995;152:1423-33.<br />
5. Greenfield S, Kaplan S, Ware JE, Jr. Exp<strong>and</strong>ing patient involvement in care. Effects on<br />
patient outcomes. Ann Intern Med. 1985;102:520-8.<br />
6. Levinson W, Roter DL, Mullooly JP, et al. Physician-patient communication. The<br />
relationship with malpractice claims among primary care physicians <strong>and</strong> surgeons.<br />
7. Lidz CW, Appelbaum PS, Meisel A. Two models <strong>of</strong> implementing informed consent. Arch<br />
Intern Med. 1988;148:1385-9.<br />
8. Kuyper AR. <strong>Patient</strong> counseling detects prescription errors. Hosp Pharm. 1993;28:1180-1,<br />
1184-9.<br />
9. Mark JS, Spiro H. Informed consent for colonoscopy. A prospective study. Arch Intern<br />
Med. 1990;150:777-80.<br />
10. Code <strong>of</strong> Federal Regulations. Title 45: <strong>Public</strong> Welfare. .Revised June 18, 1991.Effective<br />
August 19, 1991.<br />
11. Hopper KD, TenHave TR, Hartzel J. Informed consent forms for clinical <strong>and</strong> research<br />
imaging procedures: how much do patients underst<strong>and</strong>? AJR Am J Roentgenol.<br />
1995;164:493-6.<br />
12. Braddock CH, 3rd, Edwards KA, Hasenberg NM, et al. Informed decision making in<br />
outpatient practice: time to get back to basics. JAMA. 1999;282:2313-20.<br />
13. Bottrell MM, Alpert H, Fischbach RL, et al. Hospital informed consent for procedure<br />
forms: facilitating quality patient-physician interaction. Arch Surg. 2000;135:26-33.<br />
14. Boyle CM. Difference between patients' <strong>and</strong> doctors' interpretation <strong>of</strong> some common<br />
medical terms. Br Med J. 1970;1:286-9.<br />
15. Cole R. The underst<strong>and</strong>ing <strong>of</strong> medical terminology used in printed health education<br />
materials. Health Educ J. 1979;38:111-21.<br />
16. Lashley M, Talley W, L<strong>and</strong>s LC, et al. Informed proxy consent: communication between<br />
pediatric surgeons <strong>and</strong> surrogates about surgery. Pediatrics. 2000;105:591-7.<br />
17. Lavelle-Jones C, Byrne DJ, Rice P, et al. Factors affecting quality <strong>of</strong> informed consent.<br />
BMJ. 1993;306:885-90.<br />
18. Baker DW, Parker RM, Williams MV, et al. Health literacy <strong>and</strong> the risk <strong>of</strong> hospital<br />
admission. J Gen Intern Med. 1998;13:791-8.<br />
19. Kirsch I, Jungblut A, Jenkins L, et al. Adult literacy in America: a first look at the results<br />
<strong>of</strong> the national adult literacy survey. Washington, DC: National Center for Education, US<br />
Department <strong>of</strong> Education; 1993.<br />
20. Jubelirer SJ, Linton JC, Magnetti SM. Reading versus comprehension: implications for<br />
patient education <strong>and</strong> consent in an outpatient oncology clinic. J Cancer Educ. 1994;9:26-<br />
9.<br />
21. Hopper KD, Lambe HA, Shirk SJ. Readability <strong>of</strong> informed consent forms for use with<br />
iodinated contrast media. Radiology. 1993;187:279-83.<br />
553
22. Hopper KD, TenHave TR, Tully DA, et al. The readability <strong>of</strong> currently used<br />
surgical/procedure consent forms in the United States. Surgery. 1998;123:496-503.<br />
23. White CS, Mason AC, Feehan M, et al. Informed consent for percutaneous lung biopsy:<br />
comparison <strong>of</strong> two consent protocols based on patient recall after the procedure. AJR Am J<br />
Roentgenol. 1995;165:1139-42.<br />
24. Deyo RA, Cherkin DC, Weinstein J, et al. Involving patients in clinical decisions: impact<br />
<strong>of</strong> an interactive video program on use <strong>of</strong> back surgery. Med Care. 2000;38:959-69.<br />
25. Solomon J, Schwegman-Melton K. Structured teaching <strong>and</strong> patient underst<strong>and</strong>ing <strong>of</strong><br />
informed consent. Crit Care Nurse. 1987;7:74-9.<br />
26. Dawes PJ, O'Keefe L, Adcock S. Informed consent: the assessment <strong>of</strong> two structured<br />
interview approaches compared to the current approach. J Laryngol Otol. 1992; 106:420-4.<br />
27. Wadey V, Frank C. The effectiveness <strong>of</strong> patient verbalization on informed consent. Can J<br />
Surg. 1997;40:124-8.<br />
28. Hopper KD, Zajdel M, Hulse SF, et al. Interactive method <strong>of</strong> informing patients <strong>of</strong> the risks<br />
<strong>of</strong> intravenous contrast media. Radiology. 1994;192:67-71.<br />
29. Agre P, Kurtz RC, Krauss BJ. A r<strong>and</strong>omized trial using videotape to present consent<br />
information for colonoscopy. Gastrointest Endosc. 1994;40:271-6.<br />
30. Winfield AC, Ford CV, James AE, et al. Response <strong>of</strong> patients to informed consent for<br />
excretory urography. Urol Radiol. 1986;8:35-9.<br />
31. Neptune SM, Hopper KD, Houts PS, et al. Take-home informed consent for intravenous<br />
contrast media: do patients learn more? Invest Radiol. 1996;31:109-13.<br />
554
555
Chapter 49. Advance Planning For End-<strong>of</strong>-Life Care<br />
Nina Garas, MD<br />
Emory University School <strong>of</strong> Medicine<br />
Steven Z. Pantilat, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Physicians <strong>and</strong> other health care workers have long struggled with decisions regarding<br />
care for patients at the end <strong>of</strong> life. An important component <strong>of</strong> this care involves assessing <strong>and</strong><br />
underst<strong>and</strong>ing patient preferences for care through ongoing discussions with competent adult<br />
patients <strong>and</strong>/or their family members or surrogates. Advance care planning protects patient<br />
autonomy <strong>and</strong> helps to assure that their health <strong>and</strong> medical treatment wishes are implemented.<br />
Good communication at the end <strong>of</strong> life can also help patients achieve closure <strong>and</strong> meaning in the<br />
final days <strong>of</strong> their life.<br />
Over the past 20 years, public consciousness regarding planning for end-<strong>of</strong>-life care has<br />
been raised through several seminal court cases, such as those involving Karen Ann Quinlan <strong>and</strong><br />
Nancy Cruzan. These cases <strong>and</strong> the public interest they helped engender led to legislation<br />
promoting patients’ rights to determine their care at the end <strong>of</strong> life. For example, Natural Death<br />
Acts (statutes passed by State legislatures that assert a person’s right to make decisions regarding<br />
terminal care) have helped promote the use <strong>of</strong> living wills (described below). 1 In addition, in<br />
1990 the Federal <strong>Patient</strong> Self-Determination Act (PSDA) was passed by Congress to encourage<br />
competent adults to complete advance directives. The PSDA requires hospitals, nursing homes,<br />
health maintenance organizations, <strong>and</strong> hospices that participate in Medicare <strong>and</strong> Medicaid to ask<br />
if patients have advance directives, to provide information about advance directives, <strong>and</strong> to<br />
incorporate advance directives into the medical record. 2<br />
Advance directives are any expression by a patient intended to guide care, should they<br />
lose their medical decision making capacity. Although both oral <strong>and</strong> written statements are valid,<br />
the added effort required to complete written statements gives them greater weight. In addition to<br />
their use when patients lose competence, advance directives also help patients consider the type<br />
<strong>of</strong> care they would want in the future, even if they retain decision making capacity. Advance<br />
directives have legal validity in almost every State.<br />
There are 2 principal forms <strong>of</strong> written advance directives: living wills <strong>and</strong> durable powers<br />
<strong>of</strong> attorney for health care. A living will is a document that allows an individual to indicate the<br />
interventions he or she would want if he or she is terminally ill, comatose with no reasonable<br />
hope <strong>of</strong> regaining consciousness, or in a persistent vegetative state with no reasonable hope <strong>of</strong><br />
regaining significant cognitive function. A durable power <strong>of</strong> attorney for health care (DPOA-<br />
HC) is a more comprehensive document that allows an individual to appoint a person to make<br />
health care decisions for him or her should he or she lose decision making capacity.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem<br />
Respecting patient preferences regarding end-<strong>of</strong>-life care requires a well-coordinated<br />
approach. Problems can arise in both documenting patient preferences <strong>and</strong> ensuring that<br />
preferences are available <strong>and</strong> respected at the time they are needed. In addition, inadequate<br />
communication with patients can compromise the goal <strong>of</strong> respecting patient preferences for end<strong>of</strong>-life<br />
care through a variety <strong>of</strong> mechanisms.<br />
556
Failure to Document Preferences<br />
The PSDA was a legislative solution (see Chapter 55) designed to increase rates <strong>of</strong><br />
completed advance directives. Although there was initial hope that PSDA would markedly<br />
increase rates <strong>of</strong> advance directive documentation, by the early 1990s it was clear that the impact<br />
was small. At that time, a large multicenter r<strong>and</strong>omized trial, the Study to Underst<strong>and</strong> Prognoses<br />
<strong>and</strong> Preferences for Outcomes <strong>and</strong> Risks <strong>of</strong> Treatments (SUPPORT), was undertaken to improve<br />
advance care planning. SUPPORT represents one <strong>of</strong> the largest <strong>and</strong> most comprehensive efforts<br />
to describe patient preferences in seriously ill patients, <strong>and</strong> to evaluate how effectively patient<br />
preferences are communicated. SUPPORT cost 28 million dollars <strong>and</strong> enrolled 9100 seriously ill<br />
patients. In SUPPORT, a trained nurse facilitator provided prognostic information to patients <strong>and</strong><br />
medical staff, discussed patient preferences with patients <strong>and</strong> families, <strong>and</strong> facilitated<br />
communication between patients <strong>and</strong> physicians.<br />
Neither the PSDA legislation nor the SUPPORT intervention had major impacts on the<br />
documentation <strong>of</strong> patients’ preferences regarding end-<strong>of</strong>-life care. Teno et al reported on the<br />
documentation <strong>of</strong> advance directives at 3 points: before PSDA, after PSDA, <strong>and</strong> after the<br />
SUPPORT intervention. The percentage <strong>of</strong> patients with an advance directive was unchanged in<br />
all 3 groups, but documentation <strong>of</strong> those directives increased at each stage, from 6% to 35% to<br />
78% in the SUPPORT intervention group. Despite this increase in documentation, only 12% <strong>of</strong><br />
patients with an advance directive had talked with a physician when completing the document<br />
<strong>and</strong> only 25% <strong>of</strong> physicians were aware <strong>of</strong> their patients’ advance directives. 3 SUPPORT found<br />
that only 23% <strong>of</strong> seriously ill patients had talked to their doctors about their wishes concerning<br />
cardiopulmonary resuscitation (CPR) <strong>and</strong> that patient-physician discussions <strong>and</strong> decisions were<br />
uncommon even in seriously ill patients whose death was predictable. 4 Another study that<br />
surveyed elders in community settings found that the vast majority (81%) stated their desire to<br />
discuss their preferences with their physicians if they were terminally ill, but only 11% had done<br />
so. 5 As these studies demonstrate, patients <strong>of</strong>ten want to talk about death <strong>and</strong> dying but expect<br />
physicians to bring up the issues. 6<br />
Ensuring that Preferences are Available <strong>and</strong> Respected<br />
Even when advance directives are prepared, studies show they <strong>of</strong>ten do not change<br />
interventions at the end <strong>of</strong> life. 3,7 Advance directives are frequently not available, recognized or<br />
applied, nor do they help reduce hospital resource use. There are multiple reasons why advance<br />
directives may go unrecognized. 8 Admitting clerks may fail to document or incorrectly<br />
document the status <strong>of</strong> a directive on admission to the hospital. <strong>Patient</strong>s <strong>and</strong> families <strong>of</strong>ten do not<br />
inform the hospital physician or admitting clerk about their advance directives, or fail to bring<br />
documentation to the hospital. 8 In one survey <strong>of</strong> 200 patients, only 18% had filled out an<br />
advance directive <strong>and</strong> <strong>of</strong> these, 50% had secured the only copy in a safety deposit box! 9 A copy<br />
<strong>of</strong> the advance directive is <strong>of</strong>ten not transferred from the nursing home to the hospital on<br />
admission. In a study by Morrison, physicians documented advance directives or discussions<br />
with appointed proxies about treatment decisions in only 11% <strong>of</strong> admission notes. 8<br />
Although the goal <strong>of</strong> advance directives is to ensure that patients receive treatment that is<br />
consistent with their preferences, to date there is no evidence that documenting advance<br />
directives leads to this outcome. In SUPPORT, there was no evidence that increasing the rates <strong>of</strong><br />
advance directives resulted in care more consistent with patients' preferences. 10 This finding was<br />
concordant with a study <strong>of</strong> nursing home patients <strong>and</strong> their family members regarding<br />
preferences for aggressive treatment at the end <strong>of</strong> life. There, 25% <strong>of</strong> patients received care that<br />
557
was inconsistent with their previously expressed wishes. 11 The problem may not be the substance<br />
<strong>of</strong> advance directives per se, but rather in the manner in which clinicians approach them.<br />
Physicians may be hesitant to initiate discussions <strong>of</strong> advance directives with patients, especially<br />
early in the course <strong>of</strong> an illness. 12<br />
Despite these shortcomings, advance directives remain the best available approach for<br />
helping patients plan future care. These discussions, difficult as they are, help ensure that<br />
patients receive care consistent with their values <strong>and</strong> goals, spare the patient inappropriate<br />
interventions, <strong>and</strong> help maintain dignity during the dying process.<br />
Physician Communication<br />
In order to improve the quality <strong>of</strong> end-<strong>of</strong>-life care, physicians need to effectively<br />
communicate with their patients <strong>and</strong> underst<strong>and</strong> their preferences for care. Several studies have<br />
documented imperfections in physician-patient communication. 13,14 Several studies have<br />
demonstrated that physicians <strong>of</strong>ten misunderst<strong>and</strong> or are unaware <strong>of</strong> their patients’ preferences<br />
for care. 15,16 Furthermore, physician prediction <strong>of</strong> patients’ preferences for resuscitation are no<br />
better than r<strong>and</strong>om. 3,14<br />
In summary, the published literature demonstrates significant problems in all areas<br />
crucial to advance care planning <strong>and</strong> ascertainment <strong>of</strong> patient preferences, transmission <strong>of</strong><br />
information to appropriate care settings, <strong>and</strong> respecting those preferences. The provision <strong>of</strong><br />
unwanted end-<strong>of</strong>-life care is an adverse event that can potentially be avoided by the<br />
implementation <strong>of</strong> effective patient safety practices.<br />
Opportunities for Impact<br />
<strong>Patient</strong>s with chronic or life-limiting illnesses make up a large proportion <strong>of</strong> the adult<br />
primary care population. Almost three-quarters <strong>of</strong> the 2.3 million Americans that die each year<br />
are 65 years <strong>of</strong> age or older. By the year 2030, people older than 65 will compromise 20% <strong>of</strong> the<br />
total population (70 million people), compared with 13% in 1994. Today’s average life<br />
expectancy is 75.5 years, <strong>and</strong> the leading causes <strong>of</strong> death are heart disease, cancer <strong>and</strong> stroke.<br />
Data from 1995 estimated that these causes accounted for 62% <strong>of</strong> all deaths <strong>and</strong> 67% <strong>of</strong> deaths<br />
for those age 65 <strong>and</strong> over. 17 The overall picture is <strong>of</strong> an aging population, with many individuals<br />
living for several decades (<strong>of</strong>ten with chronic diseases) after the possibility <strong>of</strong> death becomes<br />
more than theoretical. 18<br />
SUPPORT documented serious problems with terminal care. Physicians did not<br />
implement patients’ refusals <strong>of</strong> interventions. When patients wished to forgo CPR, a do not<br />
resuscitate order was never written in about 50% <strong>of</strong> cases. 4 While 90% <strong>of</strong> Americans say they<br />
want to die at home, 4 out <strong>of</strong> 5 die in a hospital or other health care facility. The SUPPORT<br />
study showed that only 35% <strong>of</strong> the study patients had an advance directive. These patients had<br />
an approximate six month mortality rate <strong>of</strong> 50%.<br />
Physicians <strong>and</strong> the public also commonly overestimate the effectiveness <strong>of</strong> CPR. In<br />
reality, in-hospital cardiac arrests have a survival rate <strong>of</strong> about 15%. For patients over 65 the<br />
survival rate is about 10-11%, <strong>and</strong> 3.5% for patients over age 85. 19 Elderly nursing home patients<br />
with out-<strong>of</strong>-hospital arrest only have 1-2% survival. 20 Studies have shown that when patients are<br />
aware <strong>of</strong> the real survival rates for CPR, they are less likely to desire this intervention.<br />
558
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Documenting Preferences <strong>and</strong> Ensuring that they are Available <strong>and</strong> Respected<br />
A Physician Order form for Life-Sustaining Treatment (the POLST)<br />
In the mid-1990s, a task force <strong>of</strong> ethicists <strong>and</strong> clinicians at the Oregon Health Sciences<br />
University developed a new Do Not Resuscitate (DNR) order form called POLST (Physician<br />
Orders for Life-Sustaining Treatment). POLST is a comprehensive two-page order form that<br />
documents a patient’s preference for life-sustaining treatments. The form is designed to record a<br />
patient’s wishes clearly <strong>and</strong> simply. The accompanying POLST wallet card is included as Figure<br />
49.1; the complete POLST form <strong>and</strong> materials can be obtained from the Oregon Health Sciences<br />
University’s Center for Ethics in Health Care (http://www.ohsu.edu/ethics/polst.htm).<br />
Tolle et al examined the extent to which POLST ensured that nursing home residents’<br />
wishes were honored for DNR orders, <strong>and</strong> for hospital admission only if comfort measures<br />
failed. None <strong>of</strong> the 180 patients who completed POLST received CPR, ICU care, or ventilator<br />
support, <strong>and</strong> only 2% were hospitalized to extend life. The study subjects had low rates <strong>of</strong><br />
transfer for aggressive life-extending treatments <strong>and</strong> high levels <strong>of</strong> comfort care. 21<br />
Since 1995, more than 220,000 copies <strong>of</strong> POLST have been distributed throughout the<br />
State. Data from 1999 suggest, albeit circumstantially, that this initiative may be working. In<br />
1996, Oregon’s in-hospital mortality rate was 31%, compared with the national average <strong>of</strong><br />
56%. 22<br />
Lee et al studied the effectiveness <strong>of</strong> POLST in a Program <strong>of</strong> All-Inclusive Care for the<br />
Elderly (PACE) in Portl<strong>and</strong>, Oregon. They retrospectively reviewed POLST instructions for each<br />
<strong>of</strong> the 58 participants <strong>and</strong> whether or not each <strong>of</strong> the treatments addressed by the POLST was<br />
administered in the final 2 weeks <strong>of</strong> life. The POLST specified DNR for 50 participants (93%);<br />
CPR use was consistent with these instructions for 49 participants (91%). The participants also<br />
designated the level <strong>of</strong> care they preferred as either comfort care, limited, advanced, or full<br />
intervention. Interventions administered were at the level specified in only 25 cases (46%), with<br />
less frequent deviations in antibiotic administration, administration <strong>of</strong> IV fluids, <strong>and</strong> placement<br />
<strong>of</strong> feeding tubes. The investigators concluded that the POLST effectively limits the use <strong>of</strong> some<br />
life-sustaining interventions, but that further investigation is needed into the factors that lead<br />
physicians to deviate from patients' stated preferences about other treatments. 23<br />
559
Administrative Initiatives to Ascertain Preferences on Admission to Hospital or Nursing<br />
Home<br />
In addition to POLST, some medical centers have developed admission order forms to<br />
document patient preferences regarding end <strong>of</strong> life. These forms require health care personnel to<br />
inquire about advance directives, resuscitation preferences, artificial fluids <strong>and</strong> nutrition, etc.<br />
This approach, promoted by the passage <strong>of</strong> the PSDA, may be effective in promoting providerpatient<br />
discussions about end-<strong>of</strong>-life wishes <strong>and</strong> prevent unwanted treatments. However, there<br />
are no data documenting the effectiveness <strong>of</strong> this strategy.<br />
Ascertaining Preferences in the Outpatient Setting<br />
As with other forms <strong>of</strong> computerized decision support (Chapter 53), computer-generated<br />
reminders for primary caregivers can increase the rates <strong>of</strong> discussion <strong>of</strong> advance directives <strong>and</strong><br />
completion <strong>of</strong> advance directive forms among elderly outpatients with serious illnesses. Dexter<br />
et al performed a r<strong>and</strong>omized, controlled trial to test the effectiveness <strong>of</strong> computerized<br />
reminders. The participants were 1009 patients <strong>and</strong> 147 primary care physicians in an outpatient<br />
setting. Physicians that received computer-generated reminders that recommended discussion <strong>of</strong><br />
one or both <strong>of</strong> 2 types <strong>of</strong> advance directives were compared with physicians who received no<br />
reminders. Physicians who did not receive reminders (controls) discussed <strong>and</strong> completed<br />
advance directives in only 4% <strong>of</strong> the patients On the other h<strong>and</strong>, physicians who received both<br />
types <strong>of</strong> reminders discussed (24%) <strong>and</strong> completed (15%) advance directives significantly more<br />
frequently. 24<br />
The Portability <strong>of</strong> Advance Directives between Hospitals <strong>and</strong> Nursing Homes<br />
Ghusn et al retrospectively studied the relationship between inter-institutional<br />
communication <strong>and</strong> continuity <strong>of</strong> advance directives from hospital to nursing home settings.<br />
Having a hospital discussion about advance directives or having a hospital DNR order were<br />
associated with a higher rate <strong>of</strong> advance directive discussions in nursing homes. Hospital DNR<br />
orders were continued for 93% <strong>of</strong> patients discharged to the hospital-affiliated nursing home <strong>and</strong><br />
41% <strong>of</strong> patients discharged to the community nursing home. Specific communication <strong>of</strong> hospital<br />
DNR status to the receiving nursing homes was associated with better continuity <strong>of</strong> DNR orders.<br />
The authors concluded that completing advance directives before patients are discharged to<br />
nursing homes, communicating advance directives to the receiving home, <strong>and</strong> providing followup<br />
discussions at the nursing home might improve the continuity <strong>of</strong> advance directives between<br />
hospitals <strong>and</strong> nursing homes. 25<br />
<strong>Practices</strong> to improve physician-patient communication <strong>and</strong> physician underst<strong>and</strong>ing <strong>of</strong> patient<br />
preferences<br />
Training for Physicians<br />
Physician education is an attractive way to improve end-<strong>of</strong>-life care. Physicians <strong>of</strong>ten do<br />
not communicate about advance care planning because many have not been taught the relevant<br />
communication skills <strong>and</strong> have learned them only through personal experience. 26 A study by<br />
Tulsky et al revealed that when physicians discussed end-<strong>of</strong>-life issues with their patients, they<br />
spoke twice as much as they listened <strong>and</strong> did not routinely explore patients’ values. 26<br />
Until recently, training for health care providers in palliative care <strong>and</strong> respecting patient<br />
preferences, <strong>and</strong> materials to support such training, were inadequate. For example, recent studies<br />
560
have demonstrated that most medical <strong>and</strong> nursing textbooks insufficiently cover end-<strong>of</strong>-life care<br />
issues. 27 Increasingly, resources (including textbooks, palliative care journals or journal series, 28<br />
Web sites <strong>and</strong> training programs) are filling this educational void. The American <strong>Medical</strong><br />
Association has developed an extensive physician training program titled Education for<br />
Physicians on End-<strong>of</strong>-Life Care (EPEC). 29 This curriculum teaches fundamental skills in<br />
communication, ethical decision making, palliative care, pain <strong>and</strong> symptom management, <strong>and</strong><br />
other end-<strong>of</strong>-life treatment issues. 30 The Robert Wood Johnson Foundation initiative, “Last<br />
Acts,” is another ambitious effort to educate both patients <strong>and</strong> providers.<br />
Other educational training programs exist for physicians <strong>and</strong> students as well. 30<br />
Physicians can receive formal training by attending conferences on decisions near the end <strong>of</strong> life,<br />
case management meetings regarding individual patients, <strong>and</strong> seminars on communication skills<br />
with individual feedback to physicians on their performance. 31 Physicians with expertise in this<br />
area <strong>of</strong>ten conduct seminars to educate physicians. Buckman <strong>and</strong> Lo have developed guides for<br />
specific end-<strong>of</strong>-life discussions, such as breaking bad news <strong>and</strong> the act <strong>of</strong> active listening <strong>and</strong><br />
empathy.<br />
32, 33<br />
As attractive as these educational programs are, none have been studied for their impact<br />
on changing practice or outcomes. Although common sense might tell us that such programs are<br />
likely to be effective, the generally unimpressive relationship between pr<strong>of</strong>essional education<br />
<strong>and</strong> outcomes or process change (Chapter 54) provides grist for uncertainty pending formal<br />
effectiveness studies.<br />
Palliative Care Services<br />
Specialized palliative care programs have become increasingly common in the health<br />
care system. Physicians <strong>and</strong> other health care providers, including nurses, social workers,<br />
chaplains, <strong>and</strong> others are available to coordinate care <strong>and</strong> provide consultation for terminally ill<br />
patients in hospices, hospitals, nursing homes or patient’s homes. The palliative care service also<br />
plays an important role in fostering communication among providers, patients, <strong>and</strong> families.<br />
Data regarding effectiveness are lacking.<br />
Hospitalist Systems<br />
Hospitalist physicians may improve end-<strong>of</strong>-life care in hospitals. Hospitalists, by virtue <strong>of</strong> their<br />
large inpatient volumes, should become increasingly facile with ascertaining patient preferences<br />
regarding end-<strong>of</strong>-life care. Hospitalists have a unique opportunity to approach patients, since an<br />
admission generally signals either a worsening <strong>of</strong> the patient’s current condition or a new<br />
diagnosis. The hospitalist may have more time to spend with patients <strong>and</strong> is available over<br />
consecutive hospital days to answer any questions. A routine discussion <strong>of</strong> advance directives by<br />
hospitalists can help improve the quality <strong>and</strong> efficiency <strong>of</strong> patient care. 34 On the other h<strong>and</strong>,<br />
patients may have a long-st<strong>and</strong>ing trusting relationship with their primary care physicians, <strong>and</strong><br />
may have expressed their wishes to this physician prior to hospitalization. This possibility<br />
highlights the importance <strong>of</strong> hospitalist-primary care provider communication, particularly<br />
concerning end-<strong>of</strong>-life issues. 34<br />
One retrospective chart review study <strong>of</strong> 148 patients dying at a community teaching<br />
hospital has examined the impact <strong>of</strong> hospitalists on end-<strong>of</strong>-life care. In this study, patients cared<br />
for by hospitalists were significantly more likely to have had a documented family meeting (91%<br />
vs. 63% for patients <strong>of</strong> community-based primary physicians). About two-thirds <strong>of</strong> patients in<br />
both groups requested limitations in the level <strong>of</strong> care by the time <strong>of</strong> death. Of these, patients <strong>of</strong><br />
hospitalists were significantly less likely to have documented pain, dyspnea, or anxiety in the 48<br />
561
hours prior to death (57% vs. 75%). Whether these differences reflect differences in the quality<br />
<strong>of</strong> care, the completeness <strong>of</strong> documentation, or underlying patient differences requires further<br />
study. 35 Although the hospitalist movement holds promise for improving end-<strong>of</strong>-life discussions,<br />
more research is needed to determine whether this promise will be met.<br />
End-<strong>of</strong>-Life Education for the <strong>Public</strong><br />
Extensive public awareness <strong>and</strong> educational programs are necessary to create a foundation for<br />
successful end-<strong>of</strong>-life conversations in patients with advanced illness. Broadcasts, such as the<br />
PBS-Bill Moyers special “On Our Own Terms,” may help the public appreciate the experience<br />
<strong>of</strong> terminal illness, <strong>and</strong> the complex choices that are faced. Such presentations may encourage<br />
viewers to discuss how they might manage a similar situation, <strong>and</strong> explore their own fears <strong>and</strong><br />
concerns about dying. There are no data regarding the effectiveness <strong>of</strong> public education to<br />
improve advance care planning.<br />
Other Locally Successful Advance Care Planning Programs<br />
Individual programs to implement patient preferences have emerged around the country.<br />
Limited data suggest that they may be effective, <strong>and</strong> bear further examination as to their<br />
portability to other programs <strong>and</strong> settings <strong>and</strong> their durability over time.<br />
• “Respecting Your Choices” Program<br />
Gundersen Lutheran <strong>Medical</strong> Center in La Crosse, Wisconsin has worked on community-wide<br />
programs to improve advance care planning with an initiative called “Respecting Your Choices.”<br />
This program used patient <strong>and</strong> family education, community outreach, education for nonmedical<br />
pr<strong>of</strong>essionals, st<strong>and</strong>ard training sessions, <strong>and</strong> st<strong>and</strong>ard methods for documenting <strong>and</strong><br />
tracking advance directives. Hammes et al reported that 85% <strong>of</strong> patients in the intervention<br />
group had written advance directives at death, executed on average 1.2 years before death. Of<br />
these directives, 95% were in the medical record. Virtually all patients (95%) reported that the<br />
interview process was meaningful. The patients felt that they benefited from improved<br />
communication with loved ones <strong>and</strong> with health care providers. 36<br />
• Dayton VA Initiative<br />
The Dayton (Ohio) VA <strong>Medical</strong> Center aimed to increase the number <strong>of</strong> veterans who<br />
participated in advance care planning. VA patients <strong>and</strong> their families received a patient education<br />
booklet <strong>and</strong> a video on advance care planning. The VA also developed discussion guidelines for<br />
providers, initiated an advance care planning clinic, <strong>and</strong> initiated a bereavement support group.<br />
In a 12-week period, advance care planning discussions <strong>and</strong> follow-up increased from about 15%<br />
percent <strong>of</strong> charts to almost 90%. 37<br />
• “Let Me Decide” Program<br />
Molloy et al examined patient satisfaction with decision making <strong>and</strong> health care costs<br />
after systematically implementing an advance directive program in nursing homes. The “Let Me<br />
Decide” program included educating staff in local hospitals <strong>and</strong> nursing homes, residents, <strong>and</strong><br />
families about advance directives <strong>and</strong> <strong>of</strong>fering competent residents or next-<strong>of</strong>-kin <strong>of</strong> mentally<br />
incompetent residents an advance directive. The researchers reported that systematic<br />
implementation <strong>of</strong> this program reduced hospitalizations <strong>and</strong> aggressive care for nursing home<br />
patients who did not want that level <strong>of</strong> intervention. It also reduced utilization <strong>of</strong> health care<br />
services without affecting satisfaction or mortality. 38<br />
562
Costs <strong>and</strong> Implementation<br />
Estimating the cost <strong>of</strong> ascertaining <strong>and</strong> respecting patient preferences is difficult since<br />
improvements in this area may require major changes in the structure <strong>of</strong> the health care system.<br />
Institutional barriers, the culture <strong>of</strong> medicine, patient attitudes, time constraints physicians face<br />
with <strong>of</strong>fice visits may all play a role in implementation <strong>and</strong> may inhibit change. 26 Barriers to<br />
implementation include complacency on the part <strong>of</strong> the physician <strong>and</strong> patient, fear <strong>of</strong> political<br />
controversy, diffused responsibility, <strong>and</strong> absence (or perverse) financial incentives for providers<br />
<strong>and</strong> institutions. The surprising ineffectiveness <strong>of</strong> the SUPPORT intervention, which cost over<br />
28 million dollars, demonstrates how difficult it is to make major improvements in this area.<br />
Nevertheless, improving our ability to respect patient preferences is valuable in its own right <strong>and</strong><br />
may ultimately prove to be cost-effective, since some patients will choose to forego high<br />
technology <strong>and</strong> expensive care at the end <strong>of</strong> life.<br />
<strong>Medical</strong> care at the end <strong>of</strong> life consumes 10% to 12% <strong>of</strong> the total health care budget. An<br />
estimated 40% <strong>of</strong> the Medicare budget is spent during the last 30 days <strong>of</strong> life. 39 Some have<br />
posited that increased use <strong>of</strong> hospice <strong>and</strong> advance directives <strong>and</strong> lower use <strong>of</strong> high-technology<br />
interventions for terminally ill patients will produce significant cost savings. However, the<br />
studies on cost savings from hospice <strong>and</strong> advance directives are not definitive. The 3 r<strong>and</strong>omized<br />
trials <strong>of</strong> hospice <strong>and</strong> advance directives use show no overall savings, but the authors <strong>of</strong> a review<br />
suggest that the studies were either too small for confidence in their negative results or their<br />
intervention <strong>and</strong> cost accounting are flawed. 40 In the absence <strong>of</strong> a definitive study, the existing<br />
data suggest that hospice <strong>and</strong> advance directives can save between 25% <strong>and</strong> 40% <strong>of</strong> health care<br />
costs during the last month <strong>of</strong> life, but far less (<strong>and</strong> perhaps nothing) in the 3-12 months before<br />
death. Although, these savings are less than most people anticipate, they do indicate that hospice<br />
<strong>and</strong> advance directives should be encouraged because they certainly do not cost more <strong>and</strong> they<br />
provide a means for patients to exercise their autonomy over end-<strong>of</strong>-life decisions. 40 Finally,<br />
several <strong>of</strong> the promising interventions described above (eg, the POLST intervention), are<br />
relatively inexpensive. For example, 500 POLST forms can be ordered from Oregon Health<br />
Science University’s Web site 41 for less than $100, although the cost <strong>of</strong> implementing the<br />
POLST program is unknown.<br />
Comment<br />
Preventing unwanted aggressive care at the end <strong>of</strong> life requires active communication<br />
between provider <strong>and</strong> patient, <strong>and</strong> effective strategies to transfer information regarding<br />
preferences seamlessly across care venues. The dominant strategy to improve care in this area<br />
over the past 20 years has been the promotion <strong>of</strong> advance directives. Although the enthusiasm<br />
for advance directives has not been matched by evidence <strong>of</strong> their effectiveness, SUPPORT <strong>and</strong><br />
other studies have renewed public concern <strong>and</strong> prompted providers <strong>and</strong> policy makers to<br />
reexamine advance care planning <strong>and</strong> strive to improve it. Although we have found evidence <strong>of</strong><br />
several potentially promising strategies (perhaps the most promising <strong>of</strong> which is the POLST<br />
form), the inevitability <strong>of</strong> death <strong>and</strong> the importance patients place on improving end-<strong>of</strong>-life care<br />
point strongly to the need for further research in this area.<br />
563
Figure 49.1. POLST Wallet Card instructions<br />
POLST WALLET CARD INSTRUCTIONS<br />
This is an identification wallet card for the Physician<br />
Orders for Life-Sustaining Treatment (POLST) document.<br />
This card is not a substitute for a completed POLST<br />
document. It provides a summary <strong>of</strong> the POLST document<br />
<strong>and</strong> is expected to be honored by care providers.<br />
The POLST document <strong>and</strong> wallet card are completed by<br />
the physician. The physician must sign both the POLST<br />
document <strong>and</strong> the wallet card to make the wallet card valid.<br />
Instructions continued on other side.<br />
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _<br />
Physician Orders for Life-Sustaining Treatment<br />
Name: __________________________________________<br />
Resuscitation (<strong>Patient</strong> has no pulse <strong>and</strong> is not breathing:<br />
Resuscitate Do Not Resuscitate (DNR)<br />
<strong>Medical</strong> Interventions (has pulse <strong>and</strong>/or is breathing):<br />
Antibiotics:<br />
Comfort Measures Only Limited Interventions<br />
Advanced Interventions Full Treatment/Resuscitation<br />
No antibiotics except if needed for comfort<br />
No invasive (IM/IV) antibiotics Full Treatment<br />
Instructions continued on other side.<br />
…continued from other side<br />
It is recommended that the completed wallet card<br />
564<br />
be laminated in plastic for durability <strong>and</strong> to<br />
prevent alteration. An existing card should be<br />
destroyed if the POLST document is changed.<br />
A new wallet card can be completed to match<br />
the new physician orders.<br />
CENTER FOR ETHICS IN HEALTH CARE<br />
Oregon Health Sciences University<br />
3181 SW Sam Jackson Park Rd., UHN-86<br />
Portl<strong>and</strong>, Oregon 97201-3098<br />
(503) 494-4466<br />
_ _ _ _ _ _ __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _<br />
Artificially Administered Fluids <strong>and</strong> Nutrition:<br />
No feeding tube/IV fluids Full Treatment<br />
No long term feeding tube/IV fluids<br />
Discussed with:<br />
<strong>Patient</strong> Health Care Representative<br />
Court-appointed Guardian Spouse Other<br />
I have completed the following forms:<br />
Advance Directive Court-appointed Guardian<br />
____________________________________ ________________<br />
Print name <strong>of</strong> Physician Phone<br />
____________________________________ ________________<br />
Signature <strong>of</strong> Physician Datee<br />
Center for Ethics in Health Care Developed in conformance with ORS 127.505 et seq.<br />
©Center for Ethics in Health Care. Reproduced with permission.
References<br />
1. Jonsen AR SM, Winslade WJ. Clinical Ethics: A Practical Approach to Ethical Decisions<br />
in Clinical Medicine. Fourth ed. New York, NY: McGraw-Hill; 1998.<br />
2. Ethics manual. Fourth edition. American College <strong>of</strong> Physicians. Ann Intern Med. 1998;<br />
128: 576-94.<br />
3. Teno J, Lynn J, Wenger N, Phillips RS, Murphy DP, Connors AF, Jr., et al. Advance<br />
directives for seriously ill hospitalized patients: effectiveness with the patient selfdetermination<br />
act <strong>and</strong> the SUPPORT intervention. SUPPORT Investigators. Study to<br />
Underst<strong>and</strong> Prognoses <strong>and</strong> Preferences for Outcomes <strong>and</strong> Risks <strong>of</strong> Treatment. J Am Geriatr<br />
Soc. 1997; 45: 500-7.<br />
4. anonymous. A controlled trial to improve care for seriously ill hospitalized patients. The<br />
study to underst<strong>and</strong> prognoses <strong>and</strong> preferences for outcomes <strong>and</strong> risks <strong>of</strong> treatments<br />
(SUPPORT). The SUPPORT Principal Investigators [published erratum appears in JAMA<br />
1996 Apr 24; 275: 1232]. JAMA 1995; 274: 1591-8.<br />
5. Gamble ER, McDonald PJ, Lichstein PR. Knowledge, attitudes, <strong>and</strong> behavior <strong>of</strong> elderly<br />
persons regarding living wills. Arch Intern Med. 1991; 151: 277-80.<br />
6. Muldoon MF, Barger SD, Flory JD, Manuck SB. What are quality <strong>of</strong> life measurements<br />
measuring? BMJ. 1998; 316: 542-5.<br />
7. Hanson LC, Tulsky JA, Danis M. Can clinical interventions change care at the end <strong>of</strong> life?.<br />
Ann Intern Med. 1997; 126: 381-8.<br />
8. Morrison RS, Olson E, Mertz KR, Meier DE. The inaccessibility <strong>of</strong> advance directives on<br />
transfer from ambulatory to acute care settings. JAMA. 1995; 274: 478-82.<br />
9. Broadwell A, Boisaubin, EV, Dunn, JK, Engelhardt, HT. Advance directives on hospital<br />
admission: a survey <strong>of</strong> patient attitudes. South Med J. 1993; 86: 165-8.<br />
10. Teno JM, Licks S, Lynn J, Wenger N, Connors AF, Jr., Phillips RS, et al. Do advance<br />
directives provide instructions that direct care? SUPPORT Investigators. Study to<br />
Underst<strong>and</strong> Prognoses <strong>and</strong> Preferences for Outcomes <strong>and</strong> Risks <strong>of</strong> Treatment. J Am Geriatr<br />
Soc.1997; 45: 508-12.<br />
11. Danis M, Southerl<strong>and</strong> LI, Garrett JM, Smith JL, Hielema F, Pickard CG, et al. A<br />
prospective study <strong>of</strong> advance directives for life-sustaining care. N Engl J Med. 1991; 324:<br />
882-8.<br />
12. Christakis N, Iwashyna TJ. Attitude <strong>and</strong> self-reported practice regarding prognostication in<br />
a national sample <strong>of</strong> internists. Arch Intern Med. 1998; 158: 2389-95.<br />
13. Lo B, McLeod GA, Saika G. <strong>Patient</strong> attitudes to discussing life-sustaining treatment. Arch<br />
Intern Med. 1986; 146: 1613-5.<br />
14. anonymous. Good care <strong>of</strong> the dying patient. Council on Scientific Affairs, American<br />
<strong>Medical</strong> Association. JAMA. 1996; 275: 474-8.<br />
15. Wenger NS, Phillips RS, Teno JM, Oye RK, Dawson NV, Liu H, et al. Physician<br />
underst<strong>and</strong>ing <strong>of</strong> patient resuscitation preferences: insights <strong>and</strong> clinical implications. J Am<br />
Geriatr Soc. 2000; 48: S44-51.<br />
16. Covinsky KE, Fuller JD, Yaffe K, Johnston CB, Hamel MB, Lynn J, et al. Communication<br />
<strong>and</strong> decision-making in seriously ill patients: findings <strong>of</strong> the SUPPORT project. The Study<br />
to Underst<strong>and</strong> Prognoses <strong>and</strong> Preferences for Outcomes <strong>and</strong> Risks <strong>of</strong> Treatments. J Am<br />
Geriatr Soc. 2000; 48: S187-93.<br />
17. Approaching Death: Improving Care at the End <strong>of</strong> Life. Washington DC: National<br />
Academy Press; 1997.<br />
565
18. Donaldson M, Field, MJ. Measuring quality <strong>of</strong> care at the End <strong>of</strong> Life. Arch Intern Med.<br />
1998; 158: 121-28.<br />
19. Ebell MH, Becker LA, Barry HC, Hagen M. Survival after in-hospital cardiopulmonary<br />
resuscitation. A meta-analysis. J Gen Intern Med.1998; 13: 805-16.<br />
20. Basta L, Plunkitt K, Shassy, R, Gamouras G. Cardiopulmonary resuscitation in the elderly:<br />
Defining the limits <strong>of</strong> appropriateness. Am J Geriatr Cardiol. 1998; 7: 46-55.<br />
21. Tolle SW, Tilden VP, Nelson CA, Dunn PM. A prospective study <strong>of</strong> the efficacy <strong>of</strong> the<br />
physician order form for life-sustaining treatment. J Am Geriatr Soc. 1998; 46: 1097-102.<br />
22. Jaret P. Leading <strong>Patient</strong>s in End-<strong>of</strong>-Life Decisions. Hippocrates; 1999. p. 33-37.<br />
23. Lee M, Brummel-Smith, K, Meyer, J, Drew, N, London, MR. Physician orders for lifesustaining<br />
treatment (POLST): outcomes in a PACE program. Program <strong>of</strong> All-Inclusive<br />
Care for the Elderly. J Am Geriatr Soc. 2000; 48: 1219-25.<br />
24. Dexter P, Wolinsky, FD, Gramelspacher, GP, Zhou, ZH, Eckert, GJ, Waisburd, M,<br />
Tierney, WM. Effectiveness <strong>of</strong> computer-generated reminders for increasing discussions<br />
about advance directives <strong>and</strong> completion <strong>of</strong> advance directive forms. A r<strong>and</strong>omized,<br />
controlled trial. Ann Intern Med. 1998; 128: 102-10.<br />
25. Ghusn H, Teasdale, TA, Jordan, D. Continuity <strong>of</strong> do-not-resuscitation orders between<br />
hospital <strong>and</strong> nursing home settings. J Am Geriatr Soc. 1997; 45: 465-69.<br />
26. Tulsky JA, Fischer GS, Rose MR, Arnold RM. Opening the black box: how do physicians<br />
communicate about advance directives?. Ann Intern Med. 1998; 129: 441-9.<br />
27. Rabow M, Hardie, GE, Fair, JM, McPhee, SJ. End-<strong>of</strong>-life care content in 50 textbooks<br />
from multiple specialties. JAMA. 2000; 283: 771-8.<br />
28. McPhee S, Rabow, MW, Pantilat, SZ, Markowitz, AJ. Finding our way- - perspectives on<br />
care at the close <strong>of</strong> life. JAMA. 2000; 284: 2512-13.<br />
29. Emanuel LL, von Gunten, C.F., Ferris, F.D. The Education for Physicians on End-<strong>of</strong>-Life<br />
Care (EPEC) Curriculum; 1999.<br />
30. Larson DG, Tobin DR. End-<strong>of</strong>-life conversations: evolving practice <strong>and</strong> theory. JAMA.<br />
2000; 284: 1573-8.<br />
31. Lo B. Improving care near the end <strong>of</strong> life. Why is it so hard? [editorial; comment]. JAMA.<br />
1995; 274: 1634-6.<br />
32. Buckman R KY. How to Break Bad News: A Guide for Health Care Pr<strong>of</strong>essionals.<br />
Baltimore, MD: Johns Hopkins University Press; 1992.<br />
33. Lo B, Quill T, Tulsky J. Discussing palliative care with patients. ACP-ASIM End-<strong>of</strong>-Life<br />
Care Consensus Panel. American College <strong>of</strong> Physicians-American Society <strong>of</strong> Internal<br />
Medicine. Ann Intern Med. 1999; 130: 744-9.<br />
34. Pantilat SZ, Alpers A, Wachter RM. A new doctor in the house: ethical issues in hospitalist<br />
systems. JAMA. 1999; 282: 171-4.<br />
35. Auerbach AD, Pantilat SZ, Wachter RM, Goldman L. Processes <strong>and</strong> outcomes <strong>of</strong> end-<strong>of</strong>life<br />
care in a voluntary hospitalist model. J Gen Intern Med. 2001; 16: 115.<br />
36. Hammes BJ, Rooney BL. Death <strong>and</strong> end-<strong>of</strong>-life planning in one midwestern community.<br />
Arch Intern Med. 1998; 158: 383-90.<br />
37. Lynn J SJ, Kabcenell A. Beyond the Living Will: Advance Care Planning for All Stages <strong>of</strong><br />
Health <strong>and</strong> Disease. Improving Care for the End <strong>of</strong> Life: A Sourcebook for Health Care<br />
Managers <strong>and</strong> Clinicians: Oxford University Press; 2000. p. 73-90.<br />
38. Molloy DW, Guyatt GH, Russo R, Goeree R, O'Brien BJ, Bedard M, et al. Systematic<br />
implementation <strong>of</strong> an advance directive program in nursing homes: a r<strong>and</strong>omized<br />
controlled trial. JAMA. 2000; 283: 1437-44.<br />
566
39. Kurent J. Death <strong>and</strong> Dying in America: The need to improve End-<strong>of</strong>-Life Care. Carolina<br />
Healthcare Business 2000:16-19.<br />
40. Emanuel EJ. Cost savings at the end <strong>of</strong> life. What do the data show? JAMA. 1996; 275:<br />
1907-14.<br />
41. Center for Ethics in Health Care, Oregon Health Sciences University,<br />
http://www.ohsu.edu/ethics/polst.htm, last accessed May 8, 2001.<br />
567
568
Chapter 50. Other <strong>Practices</strong> Related to <strong>Patient</strong> Participation<br />
Laura T. Pizzi, PharmD<br />
Neil I. Goldfarb<br />
David B. Nash, MD, MBA<br />
Thomas Jefferson University School <strong>of</strong> Medicine<br />
<strong>and</strong> Office <strong>of</strong> Health Policy & Clinical Outcomes<br />
Background<br />
A number <strong>of</strong> practices <strong>and</strong> resources aim to facilitate the role <strong>of</strong> patients as their own<br />
safety advocates. These practices are not intended to shift the burden <strong>of</strong> monitoring medical error<br />
to patients. Rather, they encourage patients to share responsibility for their own safety. Although<br />
these types <strong>of</strong> interventions hold promise for enhanced patient safety, there is yet insufficient<br />
evidence <strong>of</strong> their effectiveness. Therefore, this chapter is a brief, general survey <strong>of</strong> practices<br />
related to patient participation; there are few practices that have been studied with sufficient<br />
rigor to merit a full evidence-based review. This chapter explicitly excludes consumer report<br />
cards, since such tools presently are more relevant to health care quality than patient safety. 1,2<br />
There is a substantial literature on the patient’s role in quality improvement related to specific<br />
diseases - eg, self-management <strong>and</strong> general education for patients with certain chronic diseases<br />
such as asthma, 3-5 diabetes, 6-8 <strong>and</strong> rheumatoid arthritis, 9-12 as well preoperative educational <strong>and</strong><br />
preparation programs for patients undergoing cardiac surgery. 13 This literature was not reviewed<br />
in detail, both because it falls outside our definition <strong>of</strong> patient safety practices (Chapter 1) by<br />
virtue <strong>of</strong> its disease specificity, <strong>and</strong> because the volume <strong>of</strong> material was overwhelming given the<br />
time allocated for the production <strong>of</strong> this Report. There are obvious additional opportunities to<br />
promote patient involvement in helping protect their own safety drawn from the disease-specific<br />
experiences <strong>of</strong> the past, <strong>and</strong> this should be the subject <strong>of</strong> further exposition <strong>and</strong> analysis.<br />
<strong>Patient</strong> Education Materials Regarding <strong>Patient</strong> <strong>Safety</strong><br />
Books, Web sites <strong>and</strong> consumer group publications abound with health care <strong>and</strong> medical<br />
information for patients. 14 The goal <strong>of</strong> these resources is to enable consumers to arm themselves<br />
with the knowledge to protect themselves. Health care providers may wish to distribute such<br />
materials to patients to alert them <strong>of</strong> the possible problem <strong>of</strong> medical error, <strong>and</strong> encourage those<br />
that would like to take appropriate action.<br />
The Agency for Healthcare Research <strong>and</strong> Quality 15 produces a 5-page “<strong>Patient</strong> Fact<br />
Sheet” on preventing medical errors. This fact sheet educates patients on the problem <strong>of</strong> medical<br />
error, <strong>and</strong> provides 20 tips patients may follow to avoid medical error, ranging from properly<br />
measuring liquid medications to ensuring health care employees have washed their h<strong>and</strong>s.<br />
Proprietary educational materials have also been developed. For instance, DoctorQuality,<br />
Inc., a quality management company that provides products <strong>and</strong> services to health care<br />
consumers, purchasers, <strong>and</strong> providers, has developed online <strong>and</strong> <strong>of</strong>fline tools that providers <strong>and</strong><br />
patients can use to improve care, 16 including patient safety workbooks <strong>and</strong> quality guides for a<br />
variety <strong>of</strong> diagnoses <strong>and</strong> surgical procedures. The books describe the key events that patients<br />
should anticipate at each step <strong>of</strong> diagnosis <strong>and</strong> treatment, identify high-risk points in the<br />
treatment plan where mistakes are more likely occur, 14 <strong>and</strong> provide tips as to how to avert<br />
common errors.<br />
<strong>Patient</strong>s may also find resources in the popular literature. In Lerner’s Consumer Guide to<br />
Health Care, the authors coach readers on questions to ask their physicians <strong>and</strong> ways to avoid<br />
569
medical mistakes. 17 Dr. Robert Arnot's book, The Best Medicine, 18 educates patients about<br />
specific procedures (eg, coronary artery bypass surgery, cesarean section, hysterectomy, <strong>and</strong><br />
carotid endarterectomy). Potential complications <strong>of</strong> each <strong>of</strong> these procedures are described, <strong>and</strong><br />
volumes, average lengths <strong>of</strong> stay, <strong>and</strong> complication rates <strong>of</strong> major hospitals are presented.<br />
Health information on the Web has increased patients’ desire for medical information <strong>and</strong><br />
raised significant issues regarding patient safety <strong>and</strong> the manner in which patients approach their<br />
doctors for information. 19 A recent study revealed that many physicians believe that Web<br />
resources can distance patients from physicians <strong>and</strong> have an adverse effect on patient safety.<br />
Specifically, there is concern that patients can receive <strong>and</strong> believe misinformation, or read <strong>of</strong><br />
treatments <strong>and</strong> procedures unfamiliar to physicians, <strong>and</strong> in both instances lower their trust in<br />
physician recommendations. 20 Other physicians see the Web as a positive development in patient<br />
safety because when patients approach their doctors prepared with questions, <strong>of</strong>fice visits run<br />
more smoothly, <strong>and</strong> the physician’s counsel may be better received. 20<br />
Similar issues surround the topic <strong>of</strong> direct-to-consumer (DTC) marketing by<br />
pharmaceutical companies. 21 In 1997 the Food <strong>and</strong> Drug Administration relaxed restrictions on<br />
television <strong>and</strong> advertising for prescription medication. Drug companies responded with an<br />
explosion <strong>of</strong> marketing in all forms <strong>of</strong> media. DTC advertising may stimulate discussion<br />
between patients <strong>and</strong> their doctors about treatment options, but it also drives patients to dem<strong>and</strong><br />
newer <strong>and</strong> costlier medications, when less expensive treatments might be effective. 22 When<br />
doctors resist, 46% <strong>of</strong> patients try to persuade them to change the original recommendation <strong>and</strong><br />
another 24% attempt to obtain the requested drugs from another physician. 22 These sort <strong>of</strong><br />
interactions erode the physician-patient relationship <strong>and</strong> may jeopardize safety by promoting<br />
polypharmacy.<br />
<strong>Practices</strong> to Improve Non-compliance<br />
Compliance with medical advice is widely discussed in the literature <strong>and</strong> non-compliance<br />
with treatment may result in adverse drug events. 23 The frequency <strong>of</strong> non-compliance is higher<br />
than many health care pr<strong>of</strong>essionals realize. 24 Non-compliance may arise from<br />
misunderst<strong>and</strong>ings regarding instructions for drug use, self-care, or other factors. In addition,<br />
some <strong>of</strong> these misunderst<strong>and</strong>ings may arise from remediable factors, such as language barriers 25<br />
or low health literacy. 26,27 Simple solutions, such as using a trained interpreter instead <strong>of</strong> a family<br />
member or untrained volunteer, <strong>and</strong> providing self-care <strong>and</strong> other literature in multiple languages<br />
<strong>and</strong> bilingual versions may improve patient underst<strong>and</strong>ing. Other interventions, such as patient<br />
education publications, have been proposed to reduce adverse drug events due to noncompliance.<br />
28<br />
Access to <strong>Medical</strong> Records<br />
Although patient access to their own medical records is regulated in some states, these<br />
statutes differ across the United States. 29 Some states m<strong>and</strong>ate certain levels <strong>of</strong> access for<br />
patients; others limit access or allow the provider to deem access appropriate or inappropriate.<br />
Other countries, such as Britain, have passed legislation requiring that providers allow patients to<br />
have complete access to their medical records. 30 Some argue that access to medical records may<br />
encourage patients to take a more active role in their own health care, allow patients to become<br />
better informed about their care, <strong>and</strong> increase rapport. Others argue that staff may modify<br />
medical records due to concerns about <strong>of</strong>fending the patient, <strong>and</strong> will be diverted by the time<br />
needed to explain information contained in the records. Finally, still others express concern that<br />
patients may be unnecessarily worried by what they read. 30 No studies in the United States have<br />
570
analyzed these competing views, <strong>and</strong> therefore it is not clear whether cultural norms reported in<br />
studies from other countries are applicable here, <strong>and</strong> whether allowing patients to review their<br />
own charts will have the intended effect <strong>of</strong> reducing errors.<br />
Comment<br />
With the growing level <strong>of</strong> consumerism in health care, 31 patients may wish to take a more<br />
active role in reducing their chance <strong>of</strong> experiencing a medical error. However, the research<br />
regarding the ways in which providers can facilitate this role for patients who desire it is lacking.<br />
More research is needed on the efficacy <strong>of</strong> these interventions regarding medical error reduction<br />
<strong>and</strong> on patients’ willingness <strong>and</strong> ability to use them.<br />
References<br />
1. Longo DR, L<strong>and</strong> G, Schramm W, Fraas J, Hoskins B, Howell V. Consumer reports in<br />
health care. Do they make a difference in patient care? JAMA. 1997; 278: 1579-1584.<br />
2. Marshall MN, Shekelle PG, Leatherman S, Brook RH. The public release <strong>of</strong> performance<br />
data: what do we expect to gain? A review <strong>of</strong> the evidence JAMA. 2000; 283: 1866-1874.<br />
3. Lahdensuo A, Haahtela T, Herrala J, Kava T, Kiviranta K, Kuusisto P, et al. R<strong>and</strong>omised<br />
comparison <strong>of</strong> guided self management <strong>and</strong> traditional treatment <strong>of</strong> asthma over one year.<br />
BMJ. 1996; 312: 748-752.<br />
4. Gibson PG, Coughlan J, Wilson AJ, Abramson M, Bauman A, Hensley MJ, et al. Selfmanagement<br />
education <strong>and</strong> regular practitioner review for adults with asthma. In: The<br />
Cochrane Library, Issue 2, 2001. Oxford: Update S<strong>of</strong>tware.<br />
5. Gibson PG, Coughlan J, Wilson AJ, Hensley MJ, Abramson M, Bauman A, et al. Limited<br />
(information only) patient education programs for adults with asthma. In: The Cochrane<br />
Library. 2001. Oxford: Update S<strong>of</strong>tware.<br />
6. Fain JA, Nettles A, Funnell MM, Charron D. Diabetes patient education research: an<br />
integrative literature review. Diabetes Educ. 1999; 25: 7-15.<br />
7. Norris SL, Engelgau MM, Narayan KM. Effectiveness <strong>of</strong> self-management training in type<br />
2 diabetes: a systematic review <strong>of</strong> r<strong>and</strong>omized controlled trials. Diabetes Care. 2001; 24:<br />
561-587.<br />
8. Hampson SE, Skinner TC, Hart J, Storey L, Gage H, Foxcr<strong>of</strong>t D, et al. Effects <strong>of</strong><br />
educational <strong>and</strong> psychosocial interventions for adolescents with diabetes mellitus: a<br />
systematic review. Health Technol Assess. 2001; 5: 1-79.<br />
9. Superio-Cabuslay E, Ward MM, Lorig KR. <strong>Patient</strong> education interventions in osteoarthritis<br />
<strong>and</strong> rheumatoid arthritis: a meta-analytic comparison with nonsteroidal antiinflammatory<br />
drug treatment. Arthritis Care Res. 1996; 9: 292-301.<br />
10. Brus H, van de Laar M, Taal E, Rasker J, Wiegman O. Compliance in rheumatoid arthritis<br />
<strong>and</strong> the role <strong>of</strong> formal patient education. Semin Arthritis Rheum. 1997; 26: 702-710.<br />
11. Brus HL, van de Laar MA, Taal E, Rasker JJ, Wiegman O. Effects <strong>of</strong> patient education on<br />
compliance with basic treatment regimens <strong>and</strong> health in recent onset active rheumatoid<br />
arthritis. Ann Rheum Dis. 1998; 57: 146-151.<br />
12. Barlow JH, Turner AP, Wright CC. A r<strong>and</strong>omized controlled study <strong>of</strong> the Arthritis Self-<br />
Management Programme in the UK. Health Educ Res. 2000; 15: 665-680.<br />
13. Arthur HM, Daniels C, McKelvie R, Hirsh J, Rush B. Effect <strong>of</strong> a preoperative intervention<br />
on preoperative <strong>and</strong> postoperative outcomes in low-risk patients awaiting elective coronary<br />
artery bypass graft surgery. A r<strong>and</strong>omized, controlled trial. Ann Intern Med. 2000; 133:<br />
253-262.<br />
571
14. Robinson JL, Nash DB. Consumers' role in patient safety. QRC Advis. 2000; 17: 1-3.<br />
15. <strong>Patient</strong> Fact Sheet. 20 Tips to Help Prevent <strong>Medical</strong> Errors (AHRQ <strong>Public</strong>ation No. 00-<br />
PO38) Agency for Healthcare Research <strong>and</strong> Quality. Available at:<br />
http://www.ahrq.gov/consumer/20tips.htm. Accessed June 25, 2001.<br />
16. DoctorQuality corporate website DoctorQuality.com, Inc. Available at:<br />
http://www.doctorquality.com/www/about.asp. Accessed June 25, 2001.<br />
17. Lerner P, Lerner J. Lerner's consumer guide to health care: how to get the best health care<br />
for less. Seattle, WA: Lerner Communications; 2001.<br />
18. Arnot RB. The Best Medicine: How to Choose the Top Doctors, the Top Hospitals, <strong>and</strong> the<br />
Top Treatments. Reading, MA: Addison-Wesley Publishing Company; 1993.<br />
19. Fotch E. The effect <strong>of</strong> the world wide web on you <strong>and</strong> your patients. Ophthalmol Clin<br />
North Am. 2000; 13: 261-269.<br />
20. Bard M. How will the web affect the physician-patient relationship? Interview by Marilynn<br />
Larkin. Lancet. 2000; 356: 1777.<br />
21. Wilkes MS, Bell RA, Kravitz RL. Direct-to-consumer prescription drug advertising: trends,<br />
impact, <strong>and</strong> implications. Health Aff (Millwood). 2000; 19: 110-128.<br />
22. Brown AB. The direct-to-consumer advertising dilemma. <strong>Patient</strong> Care 2001; 35:22-33.<br />
2001; 35: 22-33.<br />
23. Holt GA, Dorcheus L, Hall EL, Beck D, Ellis E, Hough J. <strong>Patient</strong> interpretation <strong>of</strong> label<br />
instructions. Am Pharm. 1992; NS32: 58-62.<br />
24. Fletcher RH. <strong>Patient</strong> compliance with therapeutic advice: a modern view. Mt Sinai J Med.<br />
1989; 56: 453-458.<br />
25. Woloshin S, Bickell NA, Schwartz LM, Gany F, Welch HG. Language barriers in medicine<br />
in the United States. JAMA. 1995; 273: 724-728.<br />
26. Williams MV, Parker RM, Baker DW, Parikh NS, Pitkin K, Coates WC, et al. Inadequate<br />
functional health literacy among patients at two public hospitals. JAMA. 1995; 274: 1677-<br />
1682.<br />
27. Gazmararian JA, Baker DW, Williams MV, Parker RM, Scott TL, Green DC, et al. Health<br />
literacy among Medicare enrollees in a managed care organization. JAMA. 1999; 281: 545-<br />
551.<br />
28. Medication safety issue brief. Asking consumers for help. Part 3. Hosp Health Netw. 2001;<br />
75: suppl 2 p. following 56.<br />
29. <strong>Patient</strong> Confidentiality American <strong>Medical</strong> Association. Available at: http://www.amaassn.org/ama/pub/category/4610.html.<br />
Accessed June 25, 2001.<br />
30. Grange A, Renvoize E, Pinder J. <strong>Patient</strong>s' rights to access their healthcare records. Nurs<br />
St<strong>and</strong>. 1998; 13: 41-42.<br />
31. Nash DB, Manfredi MP, Bozarth B, Howell S. Connecting with the new healthcare<br />
consumer. Gaithersburg, MD: Aspen Publishers, Inc.; 2001.<br />
572
PART IV. PROMOTING AND IMPLEMENTING SAFETY PRACTICES<br />
Chapter 51. Practice Guidelines<br />
Chapter 52. <strong>Critical</strong> Pathways<br />
Chapter 53. Clinical Decision Support Systems<br />
Chapter 54. Educational Techniques Used in Changing Provider Behavior<br />
Chapter 55. Legislation, Accreditation, <strong>and</strong> Market-Driven <strong>and</strong> Other Approaches to<br />
Improving <strong>Patient</strong> <strong>Safety</strong><br />
573
574
Chapter 51. Practice Guidelines<br />
Robert Trowbridge, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Scott Weingarten, MD, MPH<br />
University <strong>of</strong> California, Los Angeles School <strong>of</strong> Medicine<br />
Background<br />
Practice guidelines are among the most widely employed methods <strong>of</strong> modifying<br />
physician behavior. Defined as “systematically developed statements to assist practitioner <strong>and</strong><br />
patient decisions about appropriate health care for specific clinical conditions,” 1 guidelines may<br />
affect both the process <strong>and</strong> the outcome <strong>of</strong> care. Although guideline development <strong>and</strong><br />
implementation have traditionally focused on ensuring a perceived st<strong>and</strong>ard <strong>of</strong> care, increasing<br />
emphasis has been placed on patient outcomes <strong>and</strong> patient safety. In this regard, thous<strong>and</strong>s <strong>of</strong><br />
guidelines have been promulgated on a great variety <strong>of</strong> clinical topics ranging from the<br />
prevention <strong>of</strong> falls in the elderly to the proper use <strong>of</strong> bone marrow transplantation. 2,3<br />
Practice Description<br />
Guidelines vary greatly in terms <strong>of</strong> both method <strong>of</strong> development <strong>and</strong> format. Some<br />
consist <strong>of</strong> relatively straightforward statements that advocate a particular clinical practice,<br />
whereas others represent a series <strong>of</strong> complex algorithms that require the input <strong>of</strong> multiple clinical<br />
variables. Many guidelines are developed by specialty <strong>and</strong> advocacy organizations with attention<br />
paid to rigorously conducted systematic reviews. 4 Others may simply reflect a local st<strong>and</strong>ard <strong>of</strong><br />
care.<br />
Guidelines should only be considered a potential source <strong>of</strong> information <strong>and</strong> are effective<br />
at modifying physician behavior only when coupled with appropriate implementation strategies.<br />
Guidelines disseminated using a multifaceted approach that provides for peer influence <strong>and</strong><br />
management support, for example, are more likely to be successful in influencing provider<br />
behavior than strategies that depend solely on the passive dissemination <strong>of</strong> printed materials. 5<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> Target <strong>Safety</strong> Problem/Opportunities for Impact<br />
Practice guidelines have the potential to greatly impact patient safety as they may<br />
facilitate widespread dissemination <strong>of</strong> practices that effectively reduce medical errors. It is well<br />
established in other medical fields that known tenets <strong>of</strong> effective health care are not practiced on<br />
a universal basis despite overwhelming evidence supporting their use. According to an audit <strong>of</strong><br />
Medicaid charts in Connecticut, for example, only 50% <strong>of</strong> patients presenting with an acute<br />
myocardial infarction received aspirin <strong>and</strong> beta-blockers on the day <strong>of</strong> admission despite<br />
substantial evidence that these practices reduce mortality. 6 A second report estimated that 3500<br />
infarctions would be averted <strong>and</strong> 4300 lives saved annually if beta-blockers were appropriately<br />
prescribed in patients with coronary artery disease. 7 Guidelines could help rectify similar<br />
shortcomings in the field <strong>of</strong> patient safety with corresponding reductions in medical errors.<br />
575
Study Design<br />
There are no well-designed studies that specifically evaluate the effect <strong>of</strong> practice<br />
guidelines on patient safety. However, research regarding their effectiveness in modifying<br />
physician behavior <strong>and</strong> improving patient outcomes is plentiful. The most comprehensive review<br />
evaluating general utility <strong>of</strong> guidelines involved an extensive search <strong>of</strong> MEDLINE <strong>and</strong> other<br />
electronic databases, the gray literature <strong>and</strong> the bibliographies <strong>of</strong> pertinent articles. Its analysis<br />
included 59 reports consisting <strong>of</strong> both controlled time series <strong>and</strong> before-after studies in addition<br />
to r<strong>and</strong>omized trials. 8<br />
A second systematic review, limited to computer-based guidelines, reported the results <strong>of</strong><br />
a search <strong>of</strong> several electronic databases (MEDLINE <strong>and</strong> CINAHL) complemented by a limited<br />
bibliography search. It included 25 studies detailing the use <strong>of</strong> 20 guideline systems. In this<br />
group, there were 10 time series studies (all without external controls) <strong>and</strong> 10 controlled trials, 9<br />
<strong>of</strong> which were r<strong>and</strong>omized. 9<br />
A final systematic review, limited to the primary care setting, also searched MEDLINE<br />
<strong>and</strong> several other electronic databases <strong>and</strong> included a limited bibliography search. 10 This review<br />
included only r<strong>and</strong>omized trials that reported clinical outcomes associated with the treatment <strong>of</strong><br />
patients with established diagnoses <strong>and</strong> identified 13 studies for inclusion. In keeping with these<br />
criteria, trials <strong>of</strong> the use <strong>of</strong> guidelines to promote preventive health care <strong>and</strong> proper diagnostic<br />
evaluations were excluded.<br />
Several studies not represented in the above systematic reviews also provide valuable<br />
information on the effectiveness <strong>of</strong> practice guidelines. Two multicenter studies, for example,<br />
evaluated the impact <strong>of</strong> guidelines on the treatment <strong>of</strong> patients admitted for pneumonia <strong>and</strong><br />
certain surgical procedures, respectively. Both <strong>of</strong> these were prospective before-after trials. 11,12 A<br />
prospective controlled trial <strong>of</strong> the implementation <strong>of</strong> a diabetes guideline using problem-based<br />
learning was also completed, 13 as were 2 r<strong>and</strong>omized controlled trials <strong>of</strong> the local<br />
implementation <strong>of</strong> nationally developed guidelines. 14,15 Other recently completed works include<br />
an evaluation <strong>of</strong> a guideline for the outpatient treatment <strong>of</strong> cystitis with concurrent <strong>and</strong> historical<br />
controls, <strong>and</strong> a prospective before- after investigation <strong>of</strong> a guideline for the treatment <strong>of</strong> upper<br />
gastrointestinal hemorrhage. 16,17<br />
Study Outcomes<br />
Very few <strong>of</strong> the reviewed studies report outcomes specifically linked to patient safety.<br />
Those that do (such as guidelines to prevent falls in the elderly (Chapter 26) <strong>and</strong> to promote h<strong>and</strong><br />
washing (Chapter 12)) are <strong>of</strong> less than robust design. 2,18,23 A few <strong>of</strong> the studies included in the<br />
systematic reviews do report the effect <strong>of</strong> guidelines on surgical site infections <strong>and</strong> vaccination<br />
rates, but results are reported in terms <strong>of</strong> the process <strong>of</strong> care rather than the outcome <strong>of</strong> care. The<br />
systematic review from Worrall et al is an exception, as it only analyzed studies that reported<br />
direct clinical variables. 10 Among the other studies, however, the most commonly reported<br />
variables were length <strong>of</strong> stay, rates <strong>of</strong> provider adherence to guidelines, complication rates <strong>and</strong><br />
the rates <strong>of</strong> the appropriate use <strong>of</strong> diagnostic testing.<br />
576
Evidence for Effectiveness <strong>of</strong> Practice<br />
Despite the inherent methodologic problems <strong>of</strong> many <strong>of</strong> the cited studies, there is<br />
substantial evidence that practice guidelines may be effective in influencing provider behavior<br />
<strong>and</strong> patient outcomes. In the seminal review from Grimshaw <strong>and</strong> Russell, 9 <strong>of</strong> 11 studies<br />
reporting clinical outcomes noted some degree <strong>of</strong> improvement while the process <strong>of</strong> care was<br />
found to improve in 55 <strong>of</strong> 59 studies. 8 This review did include several trials with marginal study<br />
design, but the authors argued for the inclusion <strong>of</strong> the time series <strong>and</strong> before-after studies on the<br />
premise that the magnitude <strong>of</strong> effect seen in many <strong>of</strong> the studies overwhelmed any potential bias<br />
that may have been attributable to study design. They additionally noted that the r<strong>and</strong>omized<br />
trial may not represent the best method for evaluating practice guidelines.<br />
The systematic review investigating the utility <strong>of</strong> computer-based guidelines is similarly<br />
encouraging. 9 Fourteen <strong>of</strong> 18 studies found guidelines increased provider adherence to the tenets<br />
<strong>of</strong> care promoted by the guidelines. Of the 8 studies that evaluated patient outcomes, 3 found<br />
improvements. These results mirror those <strong>of</strong> Grimshaw <strong>and</strong> Russell <strong>and</strong> reinforce the supposition<br />
that guidelines, at least in the studies completed to date, are more effective at influencing the<br />
process rather than the outcome <strong>of</strong> care.<br />
The systematic review <strong>of</strong> controlled trials reporting clinical outcomes in the primary care<br />
setting is significantly less optimistic. 10 Only 5 <strong>of</strong> the 13 trials analyzed showed improvements in<br />
the defined clinical outcomes. None did so across all <strong>of</strong> the study groups or for an extended<br />
period <strong>of</strong> time. These results cast doubt on the utility <strong>of</strong> guidelines, particularly since this paper<br />
only included rigorously conducted trials reporting clinical outcomes, in contrast to the larger<br />
review by Grimshaw <strong>and</strong> Russell. However the authors correctly assert that many <strong>of</strong> these<br />
studies likely used guidelines that were not evidence-based <strong>and</strong> which may not have been<br />
sensitive enough to discern small improvements in clinical outcomes. In addition, it is likely that<br />
improvements in the process <strong>of</strong> care represents a reasonable surrogate endpoint for clinical<br />
outcomes given the size <strong>of</strong> the studies that would need to be conducted to show a beneficial<br />
outcomes effect <strong>of</strong> guidelines. To exclude all studies using this surrogate endpoint may be an<br />
extreme <strong>and</strong> unnecessary measure.<br />
The studies <strong>of</strong> guidelines that have been published since the completion <strong>of</strong> the above<br />
systematic reviews are similarly conflicting. The multicenter study <strong>of</strong> the pneumonia guideline,<br />
for example, showed no effect on the length <strong>of</strong> stay or complication rates whereas the study <strong>of</strong><br />
the postoperative surgical guidelines showed a significant decrease in the length <strong>of</strong> stay in 2 <strong>of</strong><br />
the 3 groups. 11,12 Additionally, studies <strong>of</strong> the implementation <strong>of</strong> nationally developed guidelines<br />
yielded conflicting results, perhaps as a function <strong>of</strong> the implementation strategies used in each<br />
study. The study <strong>of</strong> the Agency for Healthcare Policy <strong>and</strong> Research (AHCPR, now the Agency<br />
for Healthcare Research <strong>and</strong> Quality, AHRQ) guideline for cessation <strong>of</strong> smoking in pregnant<br />
woman, which was a rigorously conducted r<strong>and</strong>omized controlled trial, used extensive<br />
implementation strategies <strong>and</strong> showed a marked <strong>and</strong> statistically significant increase in the<br />
smoking cessation rate in the intervention group. 15 In contrast, the trial using continuous quality<br />
improvement measures <strong>and</strong> academic detailing to promote the AHCPR depression guideline<br />
showed little effect. 14 Yet several other studies, all <strong>of</strong> reasonable design <strong>and</strong> well-controlled,<br />
demonstrated improvements in the process <strong>of</strong> care 13,16,17 with one also reporting a marked<br />
clinical benefit in improving glycemic control in diabetics. 13<br />
577
Potential for Harm<br />
It has been theorized that practice guidelines, if improperly developed or implemented,<br />
could actually be detrimental to the process <strong>of</strong> care or worsen clinical outcomes. A study <strong>of</strong><br />
guidelines in the use <strong>of</strong> neurodiagnostic testing in patients with low back pain found the tests<br />
were more frequently utilized in an improper fashion than at baseline when clinicians were given<br />
a set <strong>of</strong> guidelines that were relatively narrow in focus. 19 Another study <strong>of</strong> patients treated for<br />
congestive heart failure found the guidelines actually increased the length <strong>of</strong> stay beyond that<br />
which was clinically necessary. 20 Although neither <strong>of</strong> these studies revealed worsened patient<br />
outcomes due to the guidelines, that potential certainly exists. In addition, the promulgation <strong>of</strong><br />
guidelines with imprudent advice could also result in widespread harm to patients.<br />
Costs <strong>and</strong> Implementation<br />
Developing <strong>and</strong> implementing practice guidelines is expensive. When developed for a<br />
complex clinical problem by a national organization, for example, they consume tremendous<br />
resources, <strong>of</strong>ten tens <strong>of</strong> thous<strong>and</strong>s <strong>of</strong> dollars. 21 In addition, some <strong>of</strong> the more successful<br />
implementation strategies, such as academic detailing (Chapter 54), also require a substantial<br />
measure <strong>of</strong> effort <strong>and</strong> financial outlay.<br />
The manner in which practice guidelines are implemented is at least as important as the<br />
content <strong>of</strong> the guidelines themselves. Several systematic reviews have investigated both the<br />
strategies that are associated with successful institution <strong>of</strong> guidelines <strong>and</strong> the specific barriers to<br />
implementation. 5, 22, 23 These reviews concluded that the deterrents vary by practice location <strong>and</strong><br />
that strategies to circumvent these barriers must be devised on an individual <strong>and</strong> local basis. In<br />
general, however, the number <strong>and</strong> intensity <strong>of</strong> the implementation strategies employed generally<br />
corresponds with the ultimate success <strong>of</strong> the guideline.<br />
Comment<br />
There is convincing but by no means overwhelming evidence that practice guidelines are<br />
effective in positively influencing the process <strong>and</strong>, to a lesser extent, the outcome <strong>of</strong> care. Many<br />
<strong>of</strong> the completed studies are plagued by methodologic shortcomings that reflect the difficulty<br />
inherent in studying the impact <strong>of</strong> guidelines. Although evidence specific to the use <strong>of</strong> guidelines<br />
in patient safety is scanty, they are likely to be as effective in this area as any other. It thus<br />
appears that well-constructed guidelines could play a significant role in ensuring patient safety<br />
<strong>and</strong> reducing medical errors. The effectiveness <strong>of</strong> guidelines, however, is dependent on many<br />
factors outside <strong>of</strong> their content. In particular, specific attention must be focused on utilizing<br />
appropriate implementation strategies if the full potential <strong>of</strong> guidelines is to be realized.<br />
References<br />
1. Field MJ LK, eds. Clinical Practice Guidelines: Directions for a new program.<br />
Washington, DC: National Academy Press; 1990.<br />
2. Baraff LJ, Lee TJ, Kader S, Della Penna R. Effect <strong>of</strong> a practice guideline for emergency<br />
department care <strong>of</strong> falls in elder patients on subsequent falls <strong>and</strong> hospitalizations for<br />
injuries. Acad Emerg Med. 1999; 6:1224-31.<br />
3. Rowe JM, Ciobanu N, Ascensao J, Stadtmauer EA, Weiner RS, Schenkein DP, et al.<br />
Recommended guidelines for the management <strong>of</strong> autologous <strong>and</strong> allogeneic bone marrow<br />
transplantation. A report from the Eastern Cooperative Oncology Group (ECOG). Ann<br />
Intern Med. 1994; 120: 143-58.<br />
578
4. Weingarten S. Using practice guideline compendiums to provide better preventive care.<br />
Ann Intern Med. 1999; 130: 454-8.<br />
5. Wensing M, van der Weijden T, Grol R. Implementing guidelines <strong>and</strong> innovations in<br />
general practice: which interventions are effective? Br J Gen Pract. 1998; 48: 991-7.<br />
6. Holmboe ES, Meehan TP, Radford MJ, Wang Y, Marciniak TA, Krumholz HM. Use <strong>of</strong><br />
critical pathways to improve the care <strong>of</strong> patients with acute myocardial infarction. Am J<br />
Med. 1999; 107: 324-31.<br />
7. Phillips KA, Shlipak MG, Coxson P, Goldman L, Heidenreich PA, Weinstein MC, et al.<br />
Underuse <strong>of</strong> beta-blockers following myocardial infarction. JAMA. 2001; 285: 1013.<br />
8. Grimshaw JM, Russell IT. Effect <strong>of</strong> clinical guidelines on medical practice: a systematic<br />
review <strong>of</strong> rigorous evaluations. Lancet. 1993; 342: 1317-22.<br />
9. Shiffman RN, Liaw Y, Br<strong>and</strong>t CA, Corb GJ. Computer-based guideline implementation<br />
systems: a systematic review <strong>of</strong> functionality <strong>and</strong> effectiveness. J Am Med Inform Assoc.<br />
1999; 6: 104-14.<br />
10. Worrall G, Chaulk P, Freake D. The effects <strong>of</strong> clinical practice guidelines on patient<br />
outcomes in primary care: a systematic review. CMAJ. 1997; 156: 1705-12.<br />
11. Weingarten S, Riedinger MS, S<strong>and</strong>hu M, Bowers C, Ellrodt AG, Nunn C, et al. Can<br />
practice guidelines safely reduce hospital length <strong>of</strong> stay? Results from a multicenter<br />
interventional study. Am J Med. 1998; 105: 33-40.<br />
12. Rhew DC, Riedinger MS, S<strong>and</strong>hu M, Bowers C, Greengold N, Weingarten SR. A<br />
prospective, multicenter study <strong>of</strong> a pneumonia practice guideline. Chest. 1998; 114: 115-9.<br />
13. Benjamin EM, Schneider MS, Hinchey KT. Implementing practice guidelines for diabetes<br />
care using problem-based learning. A prospective controlled trial using firm systems.<br />
Diabetes Care. 1999; 22: 1672-8.<br />
14. Brown JB, Shye D, McFarl<strong>and</strong> BH, Nichols GA, Mullooly JP, Johnson RE. Controlled<br />
trials <strong>of</strong> CQI <strong>and</strong> academic detailing to implement a clinical practice guideline for<br />
depression. Jt Comm J Qual Improv. 2000; 26: 39-54.<br />
15. Windsor RA, Woodby LL, Miller TM, Hardin JM, Crawford MA, DiClemente CC.<br />
Effectiveness <strong>of</strong> Agency for Health Care Policy <strong>and</strong> Research clinical practice guideline<br />
<strong>and</strong> patient education methods for pregnant smokers in medicaid maternity care. Am J<br />
Obstet Gynecol. 2000; 182: 68-75.<br />
16. Saint S, Scholes D, Fihn SD, Farrell RG, Stamm WE. The effectiveness <strong>of</strong> a clinical<br />
practice guideline for the management <strong>of</strong> presumed uncomplicated urinary tract infection<br />
in women. Am J Med. 1999; 106: 636-41.<br />
17. Hay JA ML, Weingarten SR, Ellrodt AG. Prospective evaluation <strong>of</strong> a clinical guideline<br />
recommending hospital length <strong>of</strong> stay in upper gastrointestinal tract hemorrhage. JAMA.<br />
1997: 2151-6.<br />
18. Baraff LJ, Lee TJ, Kader S, Della Penna R. Effect <strong>of</strong> a practice guideline on the process <strong>of</strong><br />
emergency department care <strong>of</strong> falls in elder patients. Acad Emerg Med. 1999; 6: 1216-23.<br />
19. Shekelle PG, Kravitz RL, Beart J, Marger M, Wang M, Lee M. Are nonspecific practice<br />
guidelines potentially harmful? A r<strong>and</strong>omized comparison <strong>of</strong> the effect <strong>of</strong> nonspecific<br />
versus specific guidelines on physician decision making. Health Serv Res. 2000; 34: 1429-<br />
48.<br />
20. Weingarten S, Riedinger M, Conner L, Johnson B, Ellrodt AG. Reducing lengths <strong>of</strong> stay in<br />
the coronary care unit with a practice guideline for patients with congestive heart failure.<br />
Insights from a controlled clinical trial. Med Care. 1994; 32: 1232-43.<br />
579
21. Smith WR. Evidence for the effectiveness <strong>of</strong> techniques to change physician behavior.<br />
Chest. 2000; 118: 8S-17S.<br />
22. Davis DA, Taylor-Vaisey A. Translating guidelines into practice. A systematic review <strong>of</strong><br />
theoretic concepts, practical experience <strong>and</strong> research evidence in the adoption <strong>of</strong> clinical<br />
practice guidelines. CMAJ. 1997; 157: 408-16.<br />
23. Cabana MD, R<strong>and</strong> CS, Powe NR, Wu AW, Wilson MH, Abboud PA, et al. Why don't<br />
physicians follow clinical practice guidelines? A framework for improvement. JAMA.<br />
1999; 282: 1458-65.<br />
580
Chapter 52. <strong>Critical</strong> Pathways<br />
Robert Trowbridge, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Scott Weingarten, MD, MPH<br />
University <strong>of</strong> California, Los Angeles School <strong>of</strong> Medicine<br />
Background<br />
Burgeoning concerns regarding patient safety, variable health care quality <strong>and</strong> increasing<br />
health care costs have led to the introduction <strong>of</strong> clinical management tools that have their origins<br />
outside <strong>of</strong> the traditional health care sector. Primary among these innovations has been the<br />
implementation <strong>of</strong> critical pathways, administrative models that streamline work <strong>and</strong> production<br />
processes. 1 <strong>Critical</strong> pathways have been utilized extensively in several different business sectors<br />
including the construction <strong>and</strong> automotives industries. 2-4 It is theorized that the adaptation <strong>of</strong><br />
pathways to health care, particularly inpatient care, may help ensure the delivery <strong>of</strong> quality care<br />
<strong>and</strong> decrease the occurrence <strong>of</strong> medical errors.<br />
Practice Description<br />
Although closely related to clinical practice guidelines (Chapter 51), pathways more<br />
directly target the specific process <strong>and</strong> sequence <strong>of</strong> care, frequently plotting out the expected<br />
course <strong>of</strong> an illness or procedure with associated prompts for appropriate interventions. Also<br />
known as clinical pathways <strong>and</strong> care maps, pathways are generally multidisciplinary by design<br />
<strong>and</strong> may incorporate the responsibilities <strong>of</strong> physicians <strong>and</strong> nurses with those <strong>of</strong> ancillary medical<br />
providers including pharmacists, physical therapists <strong>and</strong> social workers. 5 They are regularly<br />
intercalated into the point-<strong>of</strong>-care <strong>and</strong> may, in some cases, incorporate or even replace traditional<br />
chart documentation. In addition, pathways are <strong>of</strong>ten evidence-based <strong>and</strong> may even be integrated<br />
with locally or nationally developed clinical practice guidelines. Most pathways, however, are<br />
locally developed <strong>and</strong> are most frequently implemented at the level <strong>of</strong> the hospital or medical<br />
center as part <strong>of</strong> a cost-containment or quality assurance initiative.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> the Problem/Opportunities for Impact<br />
It is well established that the methods currently used to disseminate medical information<br />
are both cumbersome <strong>and</strong> inefficient. Even medical advances that are well entrenched in the<br />
literature <strong>and</strong> <strong>of</strong> unquestionable value are not routinely or universally implemented. A study <strong>of</strong><br />
patients treated for myocardial infarction in Connecticut, for example, revealed only 50% <strong>of</strong><br />
patients received the most basic beneficial treatments (aspirin <strong>and</strong> beta-blockers) at the time <strong>of</strong><br />
admission. 6 A second report suggested that 3500 myocardial infarctions would be averted <strong>and</strong><br />
4300 lives would be saved annually if all eligible patients with coronary artery disease were<br />
prescribed beta-blockers. 7 <strong>Critical</strong> pathways, if able to beneficially alter health care provider<br />
performance <strong>and</strong> ensure that effective patient safety strategies were practiced on a widespread<br />
basis, could be powerful agents in the prevention <strong>of</strong> medical errors, in addition to any beneficial<br />
impact they might have on overall health care quality.<br />
581
Study Design<br />
There is a dearth <strong>of</strong> well-designed studies analyzing the extent to which critical pathways<br />
change physician behavior <strong>and</strong> patient outcomes. Even fewer relate specifically to the topic <strong>of</strong><br />
patient safety. There are no systematic reviews <strong>of</strong> pathways <strong>and</strong> most <strong>of</strong> the published work<br />
describes the non-r<strong>and</strong>omized implementation <strong>of</strong> pathways in which it is <strong>of</strong>ten difficult to<br />
differentiate effects <strong>of</strong> the pathway from secular trends. The vast majority <strong>of</strong> these studies<br />
describe the implementation <strong>of</strong> a pathway for a specific surgical procedure <strong>and</strong> use historical<br />
controls with a retrospective before-after study design. 8-19<br />
Four r<strong>and</strong>omized controlled trials investigated the impact <strong>of</strong> the implementation <strong>of</strong><br />
pathways. 20-23 There are also several studies with non-r<strong>and</strong>omized concurrent controls with<br />
enrollment in the intervention arms being completed at the request <strong>of</strong> the attending physician. 24,25<br />
The latter group <strong>of</strong> studies also included historical control groups. There is one prospective<br />
before-after trial in which the control group actually shifted to receive the intervention after a<br />
washout period. 26 Finally one study used historical controls <strong>and</strong> concurrent controls from other<br />
hospitals in the region. 27 Table 52.1 describes the salient features <strong>of</strong> these papers.<br />
Study Outcomes<br />
The great majority <strong>of</strong> the cited studies reported at least one clinical outcome (Level 1)<br />
with the most commonly reported variable being diagnosis-related complications. Very few <strong>of</strong><br />
the studies reported more definitive endpoints, such as mortality. Most <strong>of</strong> the included studies<br />
reported surrogate clinical end-points (Level 2) as the major study outcome variables, usually<br />
length <strong>of</strong> stay <strong>and</strong> re-admission rates. The relative utilization <strong>of</strong> certain interventions including<br />
medications, laboratory tests <strong>and</strong> radiology studies was also commonly reported. In addition,<br />
most <strong>of</strong> the studies also included an analysis <strong>of</strong> the changes in costs associated with instituting<br />
the pathway. Some <strong>of</strong> the studied pathways did include other recommendations for interventions<br />
germane to the field <strong>of</strong> patient safety, including the use <strong>of</strong> indwelling urinary catheters (Chapter<br />
15) <strong>and</strong> prophylactic preoperative antibiotics (Chapter 20.1), but these outcomes were rarely<br />
reported.<br />
Evidence for Effectiveness <strong>of</strong> the Practice<br />
Several <strong>of</strong> the r<strong>and</strong>omized controlled trials provide at least some evidence that critical<br />
pathways can be effective in influencing health care provider behavior. One study evaluated the<br />
effectiveness <strong>of</strong> a pathway for the treatment <strong>of</strong> asthma in children admitted to a non-ICU<br />
hospital setting. 22 Those patients treated under the pathway, when compared to the control group<br />
undergoing “usual care,” had a shorter length <strong>of</strong> stay as well as decreased hospital charges <strong>and</strong><br />
medication usage, but no change in complication rates. Although the results are somewhat<br />
promising, the study was likely skewed by a significant Hawthorne effect, as patients treated<br />
under the pathway were placed on a separate clinical ward than those undergoing usual care<br />
(although the same physicians cared for both groups <strong>of</strong> patients). This study found no impact on<br />
the rate <strong>of</strong> complications, which provides little encouragement regarding the ability <strong>of</strong> pathways<br />
to reduce medical errors <strong>and</strong> enhance patient safety.<br />
A second r<strong>and</strong>omized trial investigated the utility <strong>of</strong> a pathway for the treatment <strong>of</strong><br />
patients undergoing knee <strong>and</strong> hip replacement at an academic medical center in Australia. 21<br />
Implementation <strong>of</strong> the pathway was followed by significant decreases in the length <strong>of</strong> stay <strong>and</strong><br />
shorter times to patient mobilization <strong>and</strong> ambulation. More importantly, it showed a decrease in<br />
the incidence <strong>of</strong> medical complications, including readmission rates (although this change did<br />
582
not reach statistical significance). The study, although fairly small at 163 total patients,<br />
represents some <strong>of</strong> the most convincing evidence that pathways may be effective in decreasing<br />
complications.<br />
A third r<strong>and</strong>omized trial, performed by the same Australian group, evaluated the effect <strong>of</strong><br />
a pathway for the treatment <strong>of</strong> patients with hip fracture. 20 Implementation <strong>of</strong> this pathway,<br />
which provided recommendations regarding medications, laboratory <strong>and</strong> radiology testing <strong>and</strong><br />
discharge planning, resulted in a significantly shorter lengths <strong>of</strong> stay without any concomitant<br />
change in complication rates. The study was rigorously conducted <strong>and</strong> complication rates were<br />
meticulously documented, but no information regarding the effort needed from clinicians to<br />
comply with the pathway was presented.<br />
A final trial used cluster r<strong>and</strong>omization to investigate the effectiveness <strong>of</strong> a critical<br />
pathway for the treatment <strong>of</strong> community-acquired pneumonia in 19 Canadian hospitals. 23 The<br />
pathway, which was initiated upon presentation <strong>of</strong> the patient to the emergency department,<br />
included recommendations for the use <strong>of</strong> a specific antibiotic <strong>and</strong> provided a clinical prediction<br />
tool to aid in decisions regarding hospital admission. Following admission, a study nurse also<br />
regularly placed guideline-based recommendations in the chart regarding changing to oral<br />
antibiotic therapy <strong>and</strong> discharge planning. The pathway showed impressive results in terms <strong>of</strong><br />
cost containment, with a shorter length <strong>of</strong> stay <strong>and</strong> a smaller percentage <strong>of</strong> inappropriate<br />
admissions. However, there were no significant changes in the clinical parameters measured,<br />
including mortality <strong>and</strong> complication rates. In addition, it is difficult to dissect the effect <strong>of</strong> the<br />
pathway from that <strong>of</strong> the other elements <strong>of</strong> the multifaceted intervention.<br />
The second strata <strong>of</strong> studies were less encouraging. A prospective before-after evaluation<br />
<strong>of</strong> the implementation <strong>of</strong> a pathway for the administration <strong>of</strong> supplemental oxygen for<br />
hospitalized patients showed the pathway was associated with markedly elevated costs without<br />
any significant clinical benefit. 26 This study, which used an intensive 3-tiered strategy including<br />
an oxygen order form, the posting <strong>of</strong> the pathway in patient rooms <strong>and</strong> a research nurse<br />
providing immediate audit <strong>and</strong> feedback, doubled the cost <strong>of</strong> providing supplemental oxygen.<br />
Although it did change some <strong>of</strong> the physician prescription practices, this alteration in practice<br />
resulted in no appreciable clinical effect.<br />
At least 2 other studies compared an intervention group with a non-r<strong>and</strong>omized<br />
concurrent control group (thus introducing the possibility <strong>of</strong> selection bias) as well as to<br />
historical controls. In a study evaluating the impact <strong>of</strong> a critical pathway on the treatment <strong>of</strong><br />
patients undergoing neck dissection, the length <strong>of</strong> stay <strong>and</strong> total costs were significantly lower<br />
for the pathway group when compared with those <strong>of</strong> the historical controls. However, the<br />
differences disappeared when the concurrent control group was used. 25 In the other, a study <strong>of</strong> a<br />
pathway for the treatment <strong>of</strong> asthma exacerbations, the pathway resulted in improvements in<br />
resource utilization that were significant compared to both historical <strong>and</strong> concurrent controls. 24<br />
However no changes were noted between the groups in readmission rates or medical outcomes.<br />
Finally, there are a great number <strong>of</strong> observational before-after studies <strong>of</strong> pathways, the<br />
vast majority <strong>of</strong> which relate to specific surgical procedures. 8-19 Few <strong>of</strong> these studies account for<br />
secular trends <strong>and</strong> their capacity for judging the effectiveness <strong>of</strong> critical pathways is limited, at<br />
best. In addition, a recent evaluation <strong>of</strong> pathways for patients undergoing a variety <strong>of</strong> surgical<br />
procedures compared the intervention group to both historical controls <strong>and</strong> to patients from<br />
similar hospitals in the same region in an attempt to correct for such secular trends. Although<br />
implementation <strong>of</strong> the pathways resulted in significant decreases in length <strong>of</strong> stay when<br />
compared to the historical controls, these differences disappeared when the pathway groups were<br />
compared to the concurrent control groups from the other hospitals. 27 An additional study <strong>of</strong> the<br />
583
effect <strong>of</strong> a pathway on the treatment <strong>of</strong> acute myocardial infarction demonstrated that the<br />
improvements seen in the intervention group were likely secondary to secular trends <strong>and</strong> not an<br />
independent effect <strong>of</strong> the pathway. 6 These powerful results cast a great deal <strong>of</strong> doubt over those<br />
studies that demonstrated effectiveness <strong>of</strong> the pathways in comparison to historical controls<br />
alone.<br />
Potential for Harm<br />
There are theoretical concerns that pathways may result in adverse patient outcomes as a<br />
result <strong>of</strong> shortened length <strong>of</strong> stay <strong>and</strong> a dependence on “cookbook medicine,” although there is<br />
little support for this in the literature.<br />
Costs <strong>and</strong> Implementation<br />
Although many studies report cost savings associated with instituting pathways, very few<br />
detail the costs <strong>of</strong> developing <strong>and</strong> implementing them. One report attempted to put a dollar figure<br />
on the development <strong>of</strong> a pathway for the treatment <strong>of</strong> patients undergoing knee replacement<br />
surgery; however the reported estimate ($21,000) did not account for the time staff physicians<br />
spent on the project. 14 The expense <strong>of</strong> developing critical pathways in terms <strong>of</strong> physician time<br />
commitment <strong>and</strong> actual financial outlay is unknown. In addition, most pathways are developed<br />
on the local level <strong>and</strong> require a great deal <strong>of</strong> initiative <strong>and</strong> expertise on the part <strong>of</strong> the hospital or<br />
medical center. Whether most centers have access to these resources is unclear.<br />
Physicians <strong>and</strong> other providers are also not universally welcoming <strong>of</strong> critical pathways.<br />
They are considered intrusive by some providers <strong>and</strong> evidence <strong>of</strong> “cookbook medicine” by<br />
others. The developers <strong>of</strong> pathways must be careful to allow for clinical judgment <strong>and</strong> flexibility<br />
in the pathways or they are likely to be ignored or applied too rigidly.<br />
Comment<br />
There is conflicting evidence regarding the efficacy <strong>of</strong> critical pathways as a method to<br />
modify health care provider behavior <strong>and</strong> a means <strong>of</strong> implementing patient safety initiatives.<br />
Although a few studies suggest they may impact physician practice <strong>and</strong>, to a lesser extent,<br />
complication rates <strong>and</strong> other clinical outcomes, the data are inconsistent <strong>and</strong> more studies are<br />
needed. Additionally it is unclear whether the costly development <strong>and</strong> implementation <strong>of</strong><br />
pathways represents an appropriate use <strong>of</strong> limited health care resources. Finally, there is very<br />
little information on the application <strong>of</strong> pathways to patient safety.<br />
584
Table 52.1. Key features <strong>of</strong> studies <strong>of</strong> critical pathways<br />
Study Setting Design,<br />
Outcomes<br />
Children with an asthma<br />
exacerbation admitted to an<br />
academic medical center: patients<br />
r<strong>and</strong>omized to pathway group or<br />
usual care group 22<br />
<strong>Patient</strong>s undergoing hip or knee<br />
replacement surgery at an academic<br />
medical center in Australia: patients<br />
r<strong>and</strong>omized to pathway group or<br />
usual care group 21<br />
<strong>Patient</strong>s being treated for hip fracture<br />
at an academic medical center in<br />
Australia 20<br />
<strong>Patient</strong>s presenting to the emergency<br />
department with pneumonia at 19<br />
medical centers across Canada:<br />
hospitals were r<strong>and</strong>omized to<br />
pathway group or usual care group 23<br />
<strong>Patient</strong>s in need <strong>of</strong> supplemental<br />
oxygen therapy: prospective beforeafter<br />
analysis with washout period 26<br />
<strong>Patient</strong>s undergoing unilateral neck<br />
dissection at an academic medical<br />
center: pathway group compared to<br />
historical controls <strong>and</strong> nonr<strong>and</strong>omized<br />
concurrent controls 25<br />
<strong>Patient</strong>s admitted to a community<br />
teaching hospital for the treatment <strong>of</strong><br />
an asthma exacerbation: pathway<br />
group compared to historical controls<br />
<strong>and</strong> non-r<strong>and</strong>omized concurrent<br />
controls 24<br />
<strong>Patient</strong>s undergoing one <strong>of</strong> several<br />
defined surgical procedures over a<br />
seven year period at an academic<br />
medical center: pathway group<br />
compared to historical controls as<br />
well as to patients from other similar<br />
hospitals in the same region 27<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 1<br />
Level 1,<br />
Level 2<br />
Level 1,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 1<br />
Level 2,<br />
Level 2<br />
585<br />
Clinical Outcomes<br />
(p value)<br />
Shorter duration <strong>of</strong><br />
intensive nebulizer<br />
treatment (0.02), length<br />
<strong>of</strong> stay (0.01)<br />
Shorter length <strong>of</strong> stay<br />
(0.011) <strong>and</strong> time to<br />
mobilization (0.001) <strong>and</strong><br />
ambulation (0.02)<br />
Shorter length <strong>of</strong> stay<br />
(0.03)<br />
Lower rates <strong>of</strong> bed-days<br />
per patient (0.04),<br />
shorter length <strong>of</strong> stay<br />
(0.01) <strong>and</strong> shorter<br />
duration <strong>of</strong> IV therapy<br />
(0.01)<br />
More frequent<br />
appropriate oxygen<br />
discontinuation orders<br />
(0.001)<br />
Length <strong>of</strong> stay decreased<br />
compared to historical<br />
controls (0.001) but not<br />
concurrent controls<br />
More appropriate<br />
antibiotic use <strong>and</strong><br />
conversion to nebulizer<br />
treatment (0.002 for<br />
historical controls, 0.05<br />
for concurrent controls)<br />
Decreased lengths <strong>of</strong><br />
stay for the pathway<br />
group compared to the<br />
historical controls, but<br />
not compared to the<br />
concurrent control group<br />
Non-Clinical<br />
Outcomes (p value)<br />
Lower room charges<br />
(0.001) <strong>and</strong> therapy<br />
charges (0.001)<br />
Increased costs (0.02)<br />
Total costs decreased<br />
compared to historical<br />
controls (0.001) but<br />
not concurrent<br />
controls
References<br />
1. Every NR, Hochman J, Becker R, Kopecky S, Cannon CP. <strong>Critical</strong> pathways : a review.<br />
Committee on Acute Cardiac Care, Council on Clinical Cardiology, American Heart<br />
Association. Circulation. 2000; 101: 461-5.<br />
2. <strong>Critical</strong> Path S<strong>of</strong>tware Smoothes Road for Automotive Supplier. Industrial Engineering.<br />
1992; 24: 28-29.<br />
3. Kallo G. The reliability <strong>of</strong> critical path method (CPM) techniques in the analysis <strong>and</strong><br />
evaluation <strong>of</strong> delay claims. Cost Engineering. 1996; 38: 35-37.<br />
4. Hatfield MN, J. The case for critical path. Cost Engineering. 1998; 40: 17-18.<br />
5. Pearson SD, Goulart-Fisher D, Lee TH. <strong>Critical</strong> pathways as a strategy for improving care:<br />
problems <strong>and</strong> potential. Ann Intern Med. 1995; 123: 941-8.<br />
6. Holmboe ES, Meehan TP, Radford MJ, Wang Y, Marciniak TA, Krumholz HM. Use <strong>of</strong><br />
critical pathways to improve the care <strong>of</strong> patients with acute myocardial infarction. Am J<br />
Med. 1999; 107: 324-31.<br />
7. Phillips KA, Shlipak MG, Coxson P, Goldman L, Heidenreich PA, Weinstein MC, et al.<br />
Underuse <strong>of</strong> beta-blockers following myocardial infarction. JAMA. 2001; 285: 1013.<br />
8. Bradshaw BG, Liu SS, Thirlby RC. St<strong>and</strong>ardized perioperative care protocols <strong>and</strong> reduced<br />
length <strong>of</strong> stay after colon surgery. J Am Coll Surg. 1998; 186: 501-6.<br />
9. Dardik A, Williams GM, Minken SL, Perler BA. Impact <strong>of</strong> a critical pathway on the results<br />
<strong>of</strong> carotid endarterectomy in a tertiary care university hospital: effect <strong>of</strong> methods on<br />
outcome. J Vasc Surg. 1997; 26: 186-92.<br />
10. Firilas AM, Higginbotham PH, Johnson DD, Jackson RJ, Wagner CW, Smith SD. A new<br />
economic benchmark for surgical treatment <strong>of</strong> appendicitis. Am Surg. 1999; 65: 769-73.<br />
11. Hanna E, Schultz S, Doctor D, Vural E, Stern S, Suen J. Development <strong>and</strong> implementation<br />
<strong>of</strong> a clinical pathway for patients undergoing total laryngectomy: impact on cost <strong>and</strong><br />
quality <strong>of</strong> care. Arch Otolaryngol Head Neck Surg. 1999; 125: 1247-51.<br />
12. Leibman BD, Dillioglugil O, Abbas F, Tanli S, Kattan MW, Scardino PT. Impact <strong>of</strong> a<br />
clinical pathway for radical retropubic prostatectomy. Urology. 1998; 52: 94-9.<br />
13. Mabrey JD, Toohey JS, Armstrong DA, Lavery L, Wammack LA. Clinical pathway<br />
management <strong>of</strong> total knee arthroplasty. Clin Orthop. 1997; 113: 125-33.<br />
14. Macario A, Horne M, Goodman S, Vitez T, Dexter F, Heinen R, et al. The effect <strong>of</strong> a<br />
perioperative clinical pathway for knee replacement surgery on hospital costs. Anesth<br />
Analg. 1998; 86: 978-84.<br />
15. March LM, Cameron ID, Cumming RG, Chamberlain AC, Schwarz JM, Brnabic AJ, et al.<br />
Mortality <strong>and</strong> morbidity after hip fracture: can evidence based clinical pathways make a<br />
difference? J Rheumatol. 2000; 27: 2227-31.<br />
16. Noedel NR, Osterloh JF, Brannan JA, Haselhorst MM, Ramage LJ, Lambrechts D. <strong>Critical</strong><br />
pathways as an effective tool to reduce cardiac transplantation hospitalization <strong>and</strong> charges.<br />
J Transpl Coord. 1996; 6: 14-9.<br />
17. Rumble SJ, Jernigan MH, Rudisill PT. Determining the effectiveness <strong>of</strong> critical pathways<br />
for coronary artery bypass graft patients: retrospective comparison <strong>of</strong> readmission rates. J<br />
Nurs Care Qual. 1996;11(2):34-40.<br />
18. Warner BW, Kulick RM, Stoops MM, Mehta S, Stephan M, Kotagal UR. An evidencedbased<br />
clinical pathway for acute appendicitis decreases hospital duration <strong>and</strong> cost. J<br />
Pediatr Surg. 1998;33(9):1371-5.<br />
586
19. Willis B, Kim LT, Anthony T, Bergen PC, Nwariaku F, Turnage RH. A clinical pathway<br />
for inguinal hernia repair reduces hospital admissions. J Surg Res. 2000;88(1):13-7.<br />
20. Choong PF, Langford AK, Dowsey MM, Santamaria NM. Clinical pathway for fractured<br />
neck <strong>of</strong> femur: a prospective, controlled study Med J Aust. 2000;172(9):423-6.<br />
21. Dowsey MM, Kilgour ML, Santamaria NM, Choong PF. Clinical pathways in hip <strong>and</strong> knee<br />
arthroplasty: a prospective r<strong>and</strong>omised controlled study. Med J Aust. 1999;170(2):59-62.<br />
22. Johnson KB, Blaisdell CJ, Walker A, Eggleston P. Effectiveness <strong>of</strong> a clinical pathway for<br />
inpatient asthma management. Pediatrics. 2000;106(5):1006-12.<br />
23. Marrie TJ, Lau CY, Wheeler SL, Wong CJ, V<strong>and</strong>ervoort MK, Feagan BG. A controlled<br />
trial <strong>of</strong> a critical pathway for treatment <strong>of</strong> community-acquired pneumonia. CAPITAL<br />
Study Investigators. Community-Acquired Pneumonia Intervention Trial Assessing<br />
Lev<strong>of</strong>loxacin. JAMA. 2000;283(6):749-55.<br />
24. Bailey R, Weingarten S, Lewis M, Mohsenifar Z. Impact <strong>of</strong> clinical pathways <strong>and</strong> practice<br />
guidelines on the management <strong>of</strong> acute exacerbations <strong>of</strong> bronchial asthma. Chest.<br />
1998;113(1):28-33.<br />
25. Chen AY, Callender D, Mansyur C, Reyna KM, Limitone E, Goepfert H. The impact <strong>of</strong><br />
clinical pathways on the practice <strong>of</strong> head <strong>and</strong> neck oncologic surgery: the University <strong>of</strong><br />
Texas M. D. Anderson Cancer Center Experience. Arch Otolaryngol Head Neck Surg.<br />
2000;126(3):322-6.<br />
26. Wong C, Visram F, Cook D, Griffith L, R<strong>and</strong>all J, O'Brien B, et al. Development,<br />
dissemination, implementation <strong>and</strong> evaluation <strong>of</strong> a clinical pathway for oxygen therapy.<br />
CMAJ. 2000;162(1):29-33.<br />
27. Pearson SD, Kleefield SF, Soukop JR, Cook EF, Lee TH. <strong>Critical</strong> pathways intervention to<br />
reduce length <strong>of</strong> hospital stay. Am J Med. 2001;110:175-80.<br />
587
588
Chapter 53. Clinical Decision Support Systems<br />
Robert Trowbridge, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Scott Weingarten, MD, MPH<br />
University <strong>of</strong> California, Los Angeles School <strong>of</strong> Medicine<br />
Background<br />
Integrating medical knowledge <strong>and</strong> advances into the clinical setting is <strong>of</strong>ten difficult due<br />
to the complexity <strong>of</strong> the involved algorithms <strong>and</strong> protocols. Clinical decision support systems<br />
(CDSS) assist the clinician in applying new information to patient care through the analysis <strong>of</strong><br />
patient-specific clinical variables. 1 Many <strong>of</strong> these systems are used to enhance diagnostic efforts<br />
<strong>and</strong> include computer-based programs such as Dxplain that provide extensive differential<br />
diagnoses based on clinical information entered by the clinician. 2 Other forms <strong>of</strong> clinical<br />
decision support systems, including antibiotic management programs <strong>and</strong> anticoagulation dosing<br />
calculators, seek to prevent medical errors <strong>and</strong> improve patient safety. 3-5<br />
Practice Description<br />
Clinical decision support systems vary greatly in their complexity, function <strong>and</strong><br />
application. 1 These clinical tools differ from practice guidelines (Chapter 51) <strong>and</strong> critical<br />
pathways (Chapter 52) in that they require the input <strong>of</strong> patient-specific clinical variables <strong>and</strong> as a<br />
result provide patient-specific recommendations. Guidelines <strong>and</strong> pathways, in contrast, may not<br />
require the input <strong>of</strong> such information <strong>and</strong> provide more general suggestions for care <strong>and</strong><br />
treatment. Although many clinical decision support systems are now computer-based, some are<br />
relatively simple, with no inherently complex internal logic systems. Among the most common<br />
forms <strong>of</strong> support systems are drug-dosing calculators, computer-based programs that calculate<br />
appropriate doses <strong>of</strong> medications after clinicians input key data (eg, patient weight, indication for<br />
drug, serum creatinine). These calculators are especially useful in managing the administration<br />
<strong>of</strong> medications with a narrow therapeutic index (see Chapter 9). More complex systems include<br />
computerized diagnostic tools that, although labor intensive <strong>and</strong> requiring extensive patientspecific<br />
data entry, may be useful as an adjunctive measure when a patient presents with a<br />
confusing constellation <strong>of</strong> symptoms <strong>and</strong> an unclear diagnosis. Other systems, both simple <strong>and</strong><br />
complex, may be integrated into the point-<strong>of</strong>-care <strong>and</strong> provide accessible reminders to clinicians<br />
regarding appropriate management based on previously entered data. These systems may be<br />
most practical when coupled with computerized physician order entry <strong>and</strong> electronic medical<br />
records (see Chapter 6). Finally, through their integration with practice guidelines <strong>and</strong> critical<br />
pathways, decision support systems may provide clinicians with suggestions for appropriate<br />
care, thus decreasing the likelihood <strong>of</strong> medical errors. For example, a guideline for the<br />
management <strong>of</strong> community-acquired pneumonia may include a clinical tool that, after the input<br />
<strong>of</strong> patient-specific data, would provide a recommendation regarding the appropriateness <strong>of</strong><br />
inpatient or outpatient therapy. 6<br />
589
Prevalence <strong>and</strong> Severity <strong>of</strong> the Target <strong>Safety</strong> Problem/Opportunities for Impact<br />
A great variety <strong>of</strong> patient safety issues may be affected by clinical decision support<br />
systems. Since the systems are typically intercalated at the point-<strong>of</strong>-care, they are likely to have<br />
their greatest effect on problems related to the individual patient. Some <strong>of</strong> the best studied<br />
applications <strong>of</strong> decision support systems are those that seek to prevent adverse drug reactions<br />
<strong>and</strong> inappropriate drug dosing as well as those that facilitate the appropriate use <strong>of</strong> effective<br />
prophylactic measures, such as those for venous thromboembolic disease (see Chapter 31). 4,5<br />
Decision support systems may also help ensure a minimum st<strong>and</strong>ard for quality <strong>of</strong> care as part <strong>of</strong><br />
the implementation <strong>of</strong> practice guidelines by providing patient-specific recommendations after<br />
the input <strong>of</strong> certain clinical variables. This combination <strong>of</strong> decision support systems <strong>and</strong><br />
guidelines may be especially effective in conjunction with the use <strong>of</strong> the electronic medical<br />
record. 7 To date no high quality studies have examined the impact <strong>of</strong> widespread institution <strong>of</strong><br />
decision support, but local studies suggest that it may be substantial. An evaluation <strong>of</strong> an<br />
antibiotic management system in a single 12-bed Intensive Care Unit over a 3 year period<br />
revealed that use <strong>of</strong> the system resulted in 24 fewer adverse drug reactions <strong>and</strong> 194 fewer cases<br />
<strong>of</strong> antibiotic-susceptibility mismatch. 4 These results <strong>and</strong> others suggest that national<br />
implementation <strong>of</strong> decision support systems could markedly improve patient safety.<br />
Study Design<br />
The functionality <strong>and</strong> effectiveness <strong>of</strong> clinical decision support systems has been<br />
evaluated in a number <strong>of</strong> systematic reviews. The seminal review, last updated in 1998,<br />
investigated the use <strong>of</strong> computer-based systems in all clinical settings. 3 The authors completed<br />
an exhaustive search <strong>of</strong> electronic databases <strong>and</strong> the bibliographies <strong>of</strong> pertinent articles <strong>and</strong><br />
found 68 prospective trials on the subject. The vast majority <strong>of</strong> these studies (90%) were<br />
r<strong>and</strong>omized trials. The articles were rated on a 10-point scale for quality. The scale assessed the<br />
design features <strong>of</strong> the trials, including method <strong>of</strong> r<strong>and</strong>omization, baseline comparability <strong>of</strong> the<br />
study groups, allocation unit, outcome measures, <strong>and</strong> degree <strong>of</strong> follow-up. The mean score on<br />
this validity scale was 6.4 for studies published prior to 1992 <strong>and</strong> 7.7 for subsequent studies.<br />
Another systematic review evaluated the use <strong>of</strong> computer-based decision aids in the<br />
provision <strong>of</strong> outpatient preventive care. 8 This study, which was based on a search <strong>of</strong> electronic<br />
databases including MEDLINE, found 16 trials that met the pre-defined inclusion criteria. Only<br />
r<strong>and</strong>omized controlled studies were included. Those that used only historical controls were<br />
excluded. The acceptable studies were then evaluated using weighted mixed effects model<br />
regression analysis.<br />
Yet another systematic review investigated the utility <strong>of</strong> computer systems in the primary<br />
care setting. 9 A detailed search <strong>of</strong> several electronic databases <strong>and</strong> a h<strong>and</strong> search <strong>of</strong><br />
bibliographies <strong>and</strong> conference proceedings yielded 30 studies that met criteria for inclusion.<br />
These papers were ranked by the same validity score utilized in the systematic review <strong>of</strong><br />
computer-based decision supports systems described above. The average validity score <strong>of</strong> the<br />
included trials was 6.7 on the 1-10 scale.<br />
Most relevant to patient safety per se is a fourth systematic review <strong>of</strong> the utility <strong>of</strong><br />
computers in medication dosing. 10 This review, which located studies in the Cochrane<br />
Collaboration on Effective Clinical Practice, was also based on an extensive search <strong>of</strong> electronic<br />
databases, supplemented by a bibliography search <strong>and</strong> consultation with experts. Of 16 relevant<br />
studies, all but one were r<strong>and</strong>omized controlled clinical trials. Time series studies were not<br />
included. The included trials were then evaluated using a r<strong>and</strong>om effects model.<br />
590
A final systematic review analyzed the utility <strong>of</strong> computers in the implementation <strong>of</strong><br />
practice guidelines which, in such a setting, may be considered clinical decision support<br />
systems. 11 A structured search <strong>of</strong> MEDLINE, CINAHL, bibliographies <strong>and</strong> books found 25<br />
papers detailing the use <strong>of</strong> 20 such systems. In this group, there were 10 time series studies (all<br />
without external controls) <strong>and</strong> 10 controlled trials, 9 <strong>of</strong> which were r<strong>and</strong>omized.<br />
Two well-designed studies have been published since the above systematic reviews. One<br />
is a cluster r<strong>and</strong>omized trial <strong>of</strong> the use <strong>of</strong> a decision support system in the treatment <strong>of</strong> patients<br />
with hypertension. 12 The second is a prospective time series investigation <strong>of</strong> a decision support<br />
system for the prevention <strong>of</strong> thromboembolic disease in post-surgical patients. 4<br />
Study Outcomes<br />
Most <strong>of</strong> the reviewed studies reported results in terms <strong>of</strong> patient outcomes <strong>and</strong> provider<br />
performance. The most common clinical outcomes (Level 2) included the relative adherence to<br />
specific recommendations <strong>and</strong> the degree to which prospectively described tenets <strong>of</strong> “appropriate<br />
practice” were followed. Examples <strong>of</strong> such principles included whether clinicians followed<br />
medication dosage recommendations <strong>and</strong> delivered appropriate preventive health care measures.<br />
Several studies reported even more definitive clinical data (Level 1) including the degree <strong>of</strong><br />
blood pressure reduction, the incidence <strong>of</strong> adverse drug reactions, the control <strong>of</strong> postoperative<br />
pain <strong>and</strong> percentage <strong>of</strong> patients with therapeutic drug blood levels.<br />
Evidence for Effectiveness <strong>of</strong> Practice<br />
The majority <strong>of</strong> the systematic reviews portray clinical decision support systems in a<br />
positive light. The seminal systematic review <strong>of</strong> computer-based decision support systems, for<br />
example, found that 43 <strong>of</strong> the 65 investigated studies showed at least some benefit in either the<br />
process <strong>of</strong> care or patient outcomes. 3 Most impressively, 74% <strong>of</strong> the studies <strong>of</strong> preventive health<br />
care reminder systems <strong>and</strong> 60% <strong>of</strong> the evaluations <strong>of</strong> drug dosing models reported a positive<br />
impact. In addition, 6 <strong>of</strong> the 14 studies that reported actual patient outcomes reported a beneficial<br />
effect. However only one <strong>of</strong> the five diagnostic aids, a model used to predict postoperative<br />
pulmonary complications, showed encouraging results. The others, including systems designed<br />
to aid diagnostic efforts in patients with chest <strong>and</strong> abdominal pain, were ineffective. This review<br />
was rigorously conducted <strong>and</strong> only those papers with robust study design were included. Overall<br />
it provides compelling evidence for the effectiveness <strong>of</strong> specific decision support systems.<br />
These results are supported by the review <strong>of</strong> systems to aid the delivery <strong>of</strong> preventive<br />
health care. 8 By applying weighted mixed effects model regression analysis to the 16 identified<br />
studies, the authors found decision support systems resulted in significantly improved rates <strong>of</strong><br />
delivery <strong>of</strong> preventive health care (OR 1.77, 95% CI: 1.38-2.27). Computer systems, however,<br />
were not statistically superior to manual reminder systems. Although this study reviewed many<br />
<strong>of</strong> the same papers that were evaluated in the reviews discussed above, it provides further<br />
evidence <strong>of</strong> the utility <strong>of</strong> these systems in improving the process <strong>of</strong> care.<br />
The review <strong>of</strong> clinical decision support systems in the outpatient setting provided<br />
similarly encouraging results. 9 All <strong>of</strong> the 21 studies that examined processes <strong>of</strong> care found<br />
improvements. Results were somewhat less impressive in the articles reporting more definitive<br />
clinical outcomes including level <strong>of</strong> blood pressure reduction <strong>and</strong> patient satisfaction. Only one<br />
<strong>of</strong> three such studies found significant benefit.<br />
The systematic review <strong>of</strong> the use <strong>of</strong> computer models to determine drug doses provides<br />
some <strong>of</strong> the most compelling evidence for application <strong>of</strong> decision support systems to patient<br />
safety <strong>and</strong> error avoidance efforts. 10 Seven <strong>of</strong> the 11 studies showed a beneficial effect on drug<br />
591
dosing, <strong>and</strong> more significantly, 4 <strong>of</strong> 6 showed a decrease in adverse drug effects. Five <strong>of</strong> six also<br />
reported direct positive effects on the patient, including one that reported improved control <strong>of</strong><br />
postoperative pain. Since the majority <strong>of</strong> the analyzed studies were r<strong>and</strong>omized controlled trials,<br />
this review provides powerful evidence that decision support systems may prevent medical<br />
errors <strong>and</strong> other adverse events. A more recent study <strong>of</strong> an antibiotic management program also<br />
supports this contention. 5 In this prospective before-after trial, the use <strong>of</strong> the decision support<br />
system resulted in a substantial <strong>and</strong> significant decrease in the incidence <strong>of</strong> adverse drug<br />
reactions (p=0.018) <strong>and</strong> prescriptions for medications contraindicated by allergy (p
Costs <strong>and</strong> Implementation<br />
Very few studies specifically address the cost <strong>of</strong> developing <strong>and</strong> implementing decision<br />
support systems. We posit that the costs may be substantial as the majority <strong>of</strong> systems are now<br />
computer-based <strong>and</strong> require significant hardware, most commonly placed at the point-<strong>of</strong>-care. In<br />
addition, many <strong>of</strong> the successful systems were integrated with computerized medical record<br />
systems <strong>and</strong> physician order entry (Chapter 6), which are not yet in universal use. The<br />
development <strong>and</strong> frequent updating <strong>of</strong> system s<strong>of</strong>tware is also likely to be very expensive.<br />
Despite these concerns, the widespread implementation <strong>of</strong> successful systems is feasible <strong>and</strong> will<br />
likely become even more so as providers <strong>and</strong> systems increasingly shift to computerized medical<br />
record systems.<br />
Comment<br />
The preponderance <strong>of</strong> evidence suggests that clinical decision support systems are at least<br />
somewhat effective. Their highest utility has been demonstrated in the prevention <strong>of</strong> medical<br />
errors, especially when coupled with a computerized medical record <strong>and</strong> directly intercalated<br />
into the care process. Unlike more passive methods such as education <strong>and</strong> feedback, decision<br />
support systems are generally able to modify physician behavior <strong>and</strong> affect the process <strong>of</strong> care.<br />
Although the results <strong>of</strong> support systems have been far less positive when used in the ongoing<br />
care <strong>of</strong> patients with chronic diseases or to help with diagnostic decision making, these<br />
capabilities may improve with further technological advances.<br />
References<br />
1. Payne TH. Computer decision support systems. Chest. 2000;118:47S-52S.<br />
2. Barnett GO, Cimino JJ, Hupp JA, H<strong>of</strong>fer EP. DXplain. An evolving diagnostic decisionsupport<br />
system. JAMA. 1987; 258: 67-74.<br />
3. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects <strong>of</strong> computer-based clinical decision<br />
support systems on physician performance <strong>and</strong> patient outcomes: a systematic review.<br />
JAMA. 1998; 280: 1339-46.<br />
4. Durieux P, Nizard R, Ravaud P, Mounier N, Lepage E. A clinical decision support system<br />
for prevention <strong>of</strong> venous thromboembolism: effect on physician behavior. JAMA. 2000;<br />
283: 2816-21.<br />
5. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF, et al. A<br />
computer-assisted management program for antibiotics <strong>and</strong> other antiinfective agents. N<br />
Engl J Med. 1998; 338: 232-8.<br />
6. Marrie TJ, Lau CY, Wheeler SL, Wong CJ, V<strong>and</strong>ervoort MK, Feagan BG. A controlled<br />
trial <strong>of</strong> a critical pathway for treatment <strong>of</strong> community-acquired pneumonia. CAPITAL<br />
Study Investigators. Community-Acquired Pneumonia Intervention Trial Assessing<br />
Lev<strong>of</strong>loxacin. JAMA. 2000; 283: 749-55.<br />
7. Shiffman RN, Br<strong>and</strong>t CA, Liaw Y, Corb GJ. A design model for computer-based guideline<br />
implementation based on information management services. J Am Med Inform Assoc.<br />
1999; 6: 99-103.<br />
8. Shea S, DuMouchel W, Bahamonde L. A meta-analysis <strong>of</strong> 16 r<strong>and</strong>omized controlled trials<br />
to evaluate computer-based clinical reminder systems for preventive care in the ambulatory<br />
setting. J Am Med Inform Assoc. 1996; 3: 399-409.<br />
9. Sullivan F, Mitchell E. Has general practitioner computing made a difference to patient<br />
care? A systematic review <strong>of</strong> published reports. BMJ. 1995; 311: 848-52.<br />
593
10. Walton R, Dovey S, Harvey E, Freemantle N. Computer support for determining drug<br />
dose: systematic review <strong>and</strong> meta-analysis. BMJ. 1999; 318: 984-90.<br />
11. Shiffman RN, Liaw Y, Br<strong>and</strong>t CA, Corb GJ. Computer-based guideline implementation<br />
systems: a systematic review <strong>of</strong> functionality <strong>and</strong> effectiveness. J Am Med Inform Assoc.<br />
1999; 6: 104-14.<br />
12. Montgomery AA, Fahey T, Peters TJ, MacIntosh C, Sharp DJ. Evaluation <strong>of</strong> computer<br />
based clinical decision support system <strong>and</strong> risk chart for management <strong>of</strong> hypertension in<br />
primary care: r<strong>and</strong>omised controlled trial. BMJ. 2000; 320: 686-90.<br />
13. Montgomery AA, Fahey T. A systematic review <strong>of</strong> the use <strong>of</strong> computers in the<br />
management <strong>of</strong> hypertension. J Epidemiol Community Health. 1998; 52: 520-5.<br />
14. Weingarten MA, Bazel D, Shannon HS. Computerized protocol for preventive medicine: a<br />
controlled self-audit in family practice. Fam Pract. 1989; 6: 120-4.<br />
594
Chapter 54. Educational Techniques Used in Changing Provider Behavior<br />
Robert Trowbridge, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Scott Weingarten, MD, MPH<br />
University <strong>of</strong> California, Los Angeles School <strong>of</strong> Medicine<br />
Background<br />
A number <strong>of</strong> techniques have been used to modify the behavior <strong>of</strong> practicing physicians. 1<br />
Continuing medical education, practice guidelines <strong>and</strong> critical pathways represent a major thrust<br />
<strong>of</strong> these efforts. The relative effectiveness <strong>of</strong> each is largely dependent on the particular strategy<br />
employed in their implementation. 2 Traditionally these strategies have focused on lectures <strong>and</strong><br />
printed materials but other techniques have also been utilized, including audit <strong>and</strong> feedback,<br />
academic detailing, local opinion leaders <strong>and</strong> reminder systems. In addition, some have<br />
championed the use <strong>of</strong> sentinel event reporting <strong>and</strong> root cause analysis in graduate medical<br />
education programs. 3 Only recently have these various techniques been critically evaluated for<br />
their effectiveness at changing physician behavior.<br />
This chapter reviews the evidence regarding the utility <strong>of</strong> educational-oriented techniques<br />
to improve provider behavior, particularly as they do or might pertain to patient safety practices.<br />
Incident reporting (Chapter 4), root cause analysis (Chapter 5), guidelines (Chapter 51),<br />
pathways (Chapter 52), <strong>and</strong> decision support systems (Chapter 53) are reviewed elsewhere in the<br />
Report.<br />
Practice Description<br />
The passive dissemination <strong>of</strong> information through the use <strong>of</strong> lectures, conferences,<br />
mailings <strong>and</strong> printed materials remains the primary method to alter physician behavior. This<br />
primacy has not been substantially challenged in practice, although more interactive techniques<br />
have increasingly been utilized. Academic detailing, for example, involves the process <strong>of</strong> having<br />
invested <strong>and</strong> well-informed agents for change interacting with individual physicians to promote<br />
certain tenets <strong>of</strong> practice. Alternatively, audit <strong>and</strong> feedback entails the review <strong>and</strong> return to the<br />
clinician <strong>of</strong> their own process <strong>of</strong> care <strong>and</strong> patient outcomes (<strong>of</strong>ten compared with local or<br />
national benchmarks or evidence-based st<strong>and</strong>ards) in the hopes it will result in more appropriate<br />
medical care. 3 Reminder systems, which may be computerized <strong>and</strong> embedded in the electronic<br />
medical record, prompt physicians to provide certain health care measures. They differ from<br />
clinical decision support systems (Chapter 53) in that they may not provide information tailored<br />
to the specific patient. Finally, opinion leaders, usually respected local physicians, may improve<br />
4, 5<br />
health care by championing “best practices” on a regional basis.<br />
Prevalence <strong>and</strong> Severity <strong>of</strong> Target <strong>Safety</strong> Problem/Opportunities for Impact<br />
It is well established that physicians are unable to keep abreast <strong>of</strong> the staggering volume<br />
<strong>of</strong> published medical literature. This is reflected by the many studies that demonstrate the glacial<br />
pace at which many beneficial advances are incorporated into medical practice. Practice<br />
guidelines, clinical decision support systems <strong>and</strong> programs for physician education are potential<br />
solutions to this problem, but their effectiveness is greatly dependent on the methods used in<br />
their implementation. 2 Despite the presence <strong>of</strong> comprehensive guidelines on the treatment <strong>of</strong><br />
reactive airways disease, for example, a substantial percentage <strong>of</strong> asthmatic patients do not<br />
595
eceive appropriate care. 6, 7 Physician education techniques that reliably impact practice patterns<br />
may yield substantial improvements in patient care <strong>and</strong> safety.<br />
Study Designs<br />
The Cochrane Group completed a series <strong>of</strong> systematic reviews <strong>of</strong> physician education<br />
based on the Research <strong>and</strong> Development Base in Continuing <strong>Medical</strong> Education, a central<br />
database compiled from an extensive search <strong>of</strong> electronic databases <strong>and</strong> bibliographies <strong>and</strong><br />
supplemented by contact with experts in the field. Although the initial review was completed in<br />
1997, reviews are regularly updated as more pertinent data are published. One such study<br />
evaluated the role <strong>of</strong> audit <strong>and</strong> feedback <strong>and</strong> found 37 r<strong>and</strong>omized controlled studies comparing<br />
this technique to non-interventional control groups. 3 An ancillary study by the Cochrane Group<br />
compared audit <strong>and</strong> feedback with other educational strategies <strong>and</strong> located 12 r<strong>and</strong>omized<br />
controlled studies for analysis. 10 A second study <strong>of</strong> the effectiveness <strong>of</strong> audit <strong>and</strong> feedback was<br />
completed by a separate group that searched MEDLINE <strong>and</strong> selected bibliographies for trials<br />
investigating the strategy’s utility in improving immunization rates. Fifteen studies were<br />
identified for inclusion, 5 <strong>of</strong> which were r<strong>and</strong>omized controlled studies with 6 interrupted time<br />
series evaluations <strong>and</strong> 4 before-after trials. 11 A third meta-analysis <strong>of</strong> peer-comparison feedback<br />
systems used an extensive electronic database <strong>and</strong> bibliography search to locate 12 r<strong>and</strong>omized<br />
controlled studies. 12<br />
The Cochrane Group also investigated the utility <strong>of</strong> academic detailing <strong>and</strong> found 18<br />
r<strong>and</strong>omized controlled studies. 13 A similar evaluation <strong>of</strong> local opinion leaders yielded 8<br />
r<strong>and</strong>omized controlled trials. 5<br />
A separate Cochrane review was completed on the utility <strong>of</strong> printed educational materials<br />
using the Cochrane Effective Practice <strong>and</strong> Organization <strong>of</strong> Care Group database. The search <strong>of</strong><br />
this database, which was compiled in the same manner as the Research <strong>and</strong> Development Base in<br />
Continuing <strong>Medical</strong> Education, found 10 r<strong>and</strong>omized controlled trials <strong>and</strong> one interrupted time<br />
series study fulfilling criteria for analysis. 14<br />
Study Outcomes<br />
Few <strong>of</strong> the studies report outcomes specific to the field <strong>of</strong> patient safety. The vast<br />
majority are concerned with process <strong>of</strong> care rather than the outcomes <strong>of</strong> care. Although clinical<br />
outcomes are reported in at least one <strong>of</strong> the studies evaluated in each <strong>of</strong> the systematic reviews<br />
(with one exception), the majority relate outcomes pertaining to physician performance. Some <strong>of</strong><br />
the more commonly described variables include the rates <strong>of</strong> appropriate provision <strong>of</strong> preventive<br />
care measures <strong>and</strong> <strong>of</strong> adherence to appropriate treatment or diagnostic protocols.<br />
Evidence for Effectiveness <strong>of</strong> Practice<br />
Much <strong>of</strong> the evidence for the effectiveness <strong>of</strong> educational <strong>and</strong> implementation techniques<br />
is <strong>of</strong> fair quality <strong>and</strong> the results are generally consistent across the various systematic reviews.<br />
However, methodologic concerns prevented the completion <strong>of</strong> quantitative data synthesis in the<br />
majority <strong>of</strong> the reviews. The studies are summarized in Table 54.1.<br />
The initial comprehensive review found overall beneficial effect for 62% <strong>of</strong><br />
interventions. In investigations <strong>of</strong> effect on patient outcomes, 48% had favorable results.<br />
Academic detailing <strong>and</strong> the use <strong>of</strong> local opinion leaders were the most effective techniques<br />
evaluated. Physician reminder systems were also effective, as 22 <strong>of</strong> the 26 evaluated studies<br />
revealed some benefit. The technique <strong>of</strong> audit <strong>and</strong> feedback was <strong>of</strong> marginal effectiveness <strong>and</strong><br />
conferences <strong>and</strong> printed materials were found to be relatively ineffective. Of note, multifaceted<br />
596
interventions with at least 3 components were associated with a 71% success rate. 8 The second<br />
comprehensive review <strong>of</strong> 102 r<strong>and</strong>omized controlled studies supported these conclusions. Yet it<br />
emphasized that the degree <strong>of</strong> effect with even the most consistently effective techniques was<br />
moderate at best, <strong>and</strong> that the process <strong>of</strong> care rather than the outcome <strong>of</strong> care was the most<br />
readily influenced variable. 9<br />
The Cochrane reviews reported similar results. Audit <strong>and</strong> feedback was found to be<br />
effective in 62% <strong>of</strong> the studies in which it was compared with non-interventional controls, but<br />
the effect was typically small. The results were not substantially different when audit <strong>and</strong><br />
feedback was augmented by conferences or educational materials or was part <strong>of</strong> a multifaceted<br />
intervention. 3 In the review <strong>of</strong> comparative trials, however, this technique was found to be<br />
inferior to reminder systems in 2 <strong>of</strong> the 3 trials where a direct comparison was made. 10 The<br />
second review <strong>of</strong> audit <strong>and</strong> feedback, which focused on improving immunization rates, found<br />
beneficial results in 4 <strong>of</strong> 5 r<strong>and</strong>omized controlled trials evaluated. Statistically significant<br />
changes were present in at least 2 <strong>of</strong> these evaluations. However, the marginal effect was small<br />
<strong>and</strong> likely was overwhelmed by the relatively high cost <strong>of</strong> the intervention. 11 Finally the metaanalysis<br />
<strong>of</strong> the 12 r<strong>and</strong>omized controlled trials investigating peer-comparison feedback systems<br />
did establish a modest benefit for the use <strong>of</strong> audit <strong>and</strong> feedback (p
Costs <strong>and</strong> Implementation<br />
Although the cost-effectiveness <strong>of</strong> the various educational techniques has not been<br />
explicitly studied, it is clear that several may require substantial outlay in terms <strong>of</strong> financial<br />
resources <strong>and</strong> personnel. It also appears that the forms <strong>of</strong> education that are most effective,<br />
including academic detailing <strong>and</strong> local opinion leaders, are also the most expensive to design <strong>and</strong><br />
support. Programs <strong>of</strong> printed materials <strong>and</strong> lectures, although dramatically less effective, are<br />
substantially less expensive to implement. It is unclear whether the integration <strong>of</strong> Internet<br />
technology <strong>and</strong> computer-based education initiatives will result in substantial changes in efficacy<br />
or cost. Finally, the relative cost-effectiveness <strong>of</strong> the various techniques remains unclear.<br />
Comment<br />
From studies <strong>of</strong> r<strong>and</strong>omized controlled trials, it appears that academic detailing <strong>and</strong> local<br />
opinion leaders are frequently associated with at least some benefit. Reminder systems are also<br />
effective in specific situations <strong>and</strong> the utility <strong>of</strong> audit <strong>and</strong> feedback has been established,<br />
although unimpressively. Traditional programs <strong>of</strong> conferences, lectures <strong>and</strong> printed materials are<br />
ineffective at inducing changes in physician behavior. None <strong>of</strong> the current techniques, however,<br />
have demonstrated a consistent ability to induce substantial <strong>and</strong> durable changes in physician<br />
behavior. The relative cost-effectiveness <strong>of</strong> the various techniques is uncertain; it remains<br />
unclear if the added cost <strong>of</strong> the more effective strategies (ie, academic detailing <strong>and</strong> local opinion<br />
leaders) is justified given their relatively small marginal increase in effectiveness. Finally there<br />
are few data regarding the specific utility <strong>of</strong> these techniques in increasing patient safety <strong>and</strong>/or<br />
the prevention <strong>of</strong> medical errors. However, techniques effective in other areas <strong>of</strong> medicine are<br />
likely to be equally effective in inducing practices changes to improve patient safety.<br />
598
Table 54.1. Studies <strong>of</strong> techniques for changing physician behavior*<br />
Study Setting Study<br />
Design<br />
Review <strong>of</strong> 99 trials assessing the<br />
effect <strong>of</strong> educational techniques on<br />
physician performance in all clinical<br />
settings 8<br />
Review <strong>of</strong> 102 trials assessing the<br />
effect <strong>of</strong> educational techniques on<br />
physician performance in all clinical<br />
settings 9<br />
Review <strong>of</strong> 37 r<strong>and</strong>omized controlled<br />
trials <strong>of</strong> the utility <strong>of</strong> audit <strong>and</strong><br />
feedback in all clinical settings in the<br />
US, Europe <strong>and</strong> Australia 3<br />
Review <strong>of</strong> 12 trials comparing the<br />
effect <strong>of</strong> audit <strong>and</strong> feedback with<br />
other educational techniques on 2194<br />
physicians in all clinical settings 10<br />
Review <strong>of</strong> fifteen studies, <strong>of</strong> which<br />
five were r<strong>and</strong>omized controlled<br />
trials, investigating the use <strong>of</strong> audit<br />
<strong>and</strong> feedback in improving<br />
immunization rates in adults <strong>and</strong><br />
children in the outpatient setting in<br />
the US <strong>and</strong> the UK 11<br />
Review <strong>of</strong> 18 trials investigating the<br />
effects <strong>of</strong> academic detailing on 1896<br />
physicians in the US, Canada,<br />
Europe, Indonesia <strong>and</strong> Australia 13<br />
Review <strong>of</strong> 8 r<strong>and</strong>omized controlled<br />
trials investigating the effect <strong>of</strong> local<br />
opinion leaders on 296 physicians in<br />
the US, Canada <strong>and</strong> Hong Kong 5<br />
Review <strong>of</strong> eleven studies evaluating<br />
the effect <strong>of</strong> printed education<br />
materials on over 1848 physicians in<br />
a variety <strong>of</strong> clinical settings 14<br />
599<br />
Results<br />
Level 1A 62% <strong>of</strong> the interventions were associated with<br />
beneficial results; academic detailing, local opinion<br />
leaders <strong>and</strong> reminder systems were the most<br />
effective while audit <strong>and</strong> feedback was less so;<br />
traditional CME programs were ineffective<br />
Level 1A Academic detailing <strong>and</strong> local opinion leaders were<br />
the most effective techniques; audit <strong>and</strong> feedback<br />
<strong>and</strong> reminder systems were less effective;<br />
multifaceted approaches were effective, especially<br />
at influencing the process <strong>of</strong> care<br />
Level 1A Eight <strong>of</strong> 13 studies showed a moderate beneficial<br />
effect with audit <strong>and</strong> feedback with little change<br />
noted when other interventions were added or a<br />
multifaceted approach was used<br />
Level 1A Two <strong>of</strong> 3 trials showed reminder systems<br />
outperformed audit <strong>and</strong> feedback; 4 studies<br />
demonstrated little benefit to adding other<br />
modalities to audit <strong>and</strong> feedback<br />
Level 1A Twelve <strong>of</strong> the 15 studies showed a benefit with<br />
audit <strong>and</strong> feedback <strong>and</strong> <strong>of</strong> the 5 RCTs, 4 showed a<br />
beneficial trend that was significant in at least 2 <strong>of</strong><br />
the trials<br />
Level 1A All <strong>of</strong> the evaluated studies showed some degree <strong>of</strong><br />
benefit although only one looked specifically at<br />
patient outcomes<br />
Level 1A Six <strong>of</strong> the 7 studies evaluating effects on physician<br />
performance showed a beneficial effect with 2 <strong>of</strong><br />
these being statistically significant; one <strong>of</strong> the 3<br />
trials evaluating patient outcomes showed a<br />
significantly positive effect<br />
Level 1A None <strong>of</strong> the studies reported significantly improved<br />
outcomes with the use <strong>of</strong> printed educational<br />
materials<br />
* CME indicates continuing medical education; RCT, r<strong>and</strong>omized controlled trial.
References<br />
1. Smith WR. Evidence for the effectiveness <strong>of</strong> techniques To change physician behavior.<br />
Chest. 2000; 118: 8S-17S.<br />
2. Grimshaw JM, Russell IT. Effect <strong>of</strong> clinical guidelines on medical practice: a systematic<br />
review <strong>of</strong> rigorous evaluations. Lancet. 1993; 342: 1317-22.<br />
3. Thomson O'Brien MA, Oxman AD, Davis DA, Haynes RB, Freemantle N, Harvey EL.<br />
Audit <strong>and</strong> feedback: effects on pr<strong>of</strong>essional practice <strong>and</strong> health care outcomes. In: The<br />
Cochrane Library, Issue 2, 2000:Oxford: Update S<strong>of</strong>tware.<br />
4. Szilagyi PG, Bordley C, Vann JC, Chelminski A, Kraus RM, Margolis PA, et al. Effect <strong>of</strong><br />
patient reminder/recall interventions on immunization rates: A review. JAMA. 2000; 284:<br />
1820-7.<br />
5. Thomson O'Brien MA, Oxman AD, Haynes RB, Davis DA, Freemantle N, Harvey EL.<br />
Local opinion leaders: effects on pr<strong>of</strong>essional practice <strong>and</strong> health care outcomes. In: The<br />
Cochrane Library, Issue 2, 2000. Oxford: Update S<strong>of</strong>tware..<br />
6. Apter AJ, Van Ho<strong>of</strong> TJ, Sherwin TE, Casey BA, Petrillo MK, Meehan TP. Assessing the<br />
quality <strong>of</strong> asthma care provided to Medicaid patients enrolled in managed care<br />
organizations in Connecticut. Ann Allergy Asthma Immunol. 2001; 86: 211-8.<br />
7. Rabe KF, Vermeire PA, Soriano JB, Maier WC. Clinical management <strong>of</strong> asthma in 1999:<br />
the Asthma Insights <strong>and</strong> Reality in Europe (AIRE) study. Eur Respir J. 2000; 16: 802-7.<br />
8. Davis DA, Thomson MA, Oxman AD, Haynes RB. Changing physician performance. A<br />
systematic review <strong>of</strong> the effect <strong>of</strong> continuing medical education strategies. JAMA. 1995;<br />
274: 700-5.<br />
9. Oxman AD, Thomson MA, Davis DA, Haynes RB. No magic bullets: a systematic review<br />
<strong>of</strong> 102 trials <strong>of</strong> interventions to improve pr<strong>of</strong>essional practice. CMAJ. 1995; 153: 1423-31.<br />
10. Thomson O'Brien MA, Oxman AD, Davis DA, Haynes RB, Freemantle N, Harvey EL.<br />
Audit <strong>and</strong> feedback versus alternative strategies: effects on pr<strong>of</strong>essional practice <strong>and</strong> health<br />
care outcomes. In: The Cochrane Library, Issue 2, 2000. Oxford: Update S<strong>of</strong>tware.<br />
11. Bordley WC, Chelminski A, Margolis PA, Kraus R, Szilagyi PG, Vann JJ. The effect <strong>of</strong><br />
audit <strong>and</strong> feedback on immunization delivery: a systematic review. Am J Prev Med. 2000;<br />
18: 343-50.<br />
12. Balas EA, Boren SA, Brown GD, Ewigman BG, Mitchell JA, Perk<strong>of</strong>f GT. Effect <strong>of</strong><br />
physician pr<strong>of</strong>iling on utilization. Meta-analysis <strong>of</strong> r<strong>and</strong>omized clinical trials. J Gen Intern<br />
Med. 1996; 11: 584-90.<br />
13. Thomson O'Brien MA, Oxman AD, Davis DA, Haynes RB, Freemantle N, Harvey EL.<br />
Educational outreach visits: effects on pr<strong>of</strong>essional practice <strong>and</strong> health care outcomes. In:<br />
The Cochrane Library, Issue 2, 2000. Oxford: Update S<strong>of</strong>tware.<br />
14. Freemantle N, Harvey EL, Wolf F, Grimshaw JM, Grilli R, Bero LA. Printed educational<br />
materials: effects on pr<strong>of</strong>essional practice <strong>and</strong> health care outcomes. In: The Cochrane<br />
Library, Issue 2, 2000. Oxford: Update S<strong>of</strong>tware.<br />
600
Chapter 55. Legislation, Accreditation, <strong>and</strong> Market-Driven <strong>and</strong> Other<br />
Approaches to Improving <strong>Patient</strong> <strong>Safety</strong><br />
Robert Trowbridge, MD<br />
Robert M.Wachter, MD<br />
University <strong>of</strong> California, San Francisco School <strong>of</strong> Medicine<br />
Background<br />
Although conventional approaches to health care quality improvement based on<br />
educating providers <strong>and</strong> <strong>of</strong>fering feedback have been proposed, the tremendous public <strong>and</strong><br />
pr<strong>of</strong>essional concerns raised by the Institute <strong>of</strong> Medicine’s report on medical errors 1 have led to<br />
an unusual amount <strong>of</strong> interest in regulatory <strong>and</strong> legislative efforts to improve safety. Given the<br />
traditional American dependence on education, competition, <strong>and</strong> other non-coercive mechanisms<br />
<strong>of</strong> change, the shift toward a regulatory approach is evidence <strong>of</strong> the depth <strong>of</strong> concern this issue<br />
has engendered. As these regulatory <strong>and</strong> legislative approaches are considered, it is also<br />
worthwhile to consider the role <strong>of</strong> health care payers, hospital accreditation organizations, <strong>and</strong><br />
pr<strong>of</strong>essional societies, all <strong>of</strong> whom have also led safety initiatives. 2-4 This chapter considers the<br />
potential advantages <strong>and</strong> disadvantages <strong>of</strong> legislative, regulatory, pr<strong>of</strong>essional society, <strong>and</strong><br />
market-oriented approaches to implementing patient safety efforts, <strong>and</strong> reviews the evidence<br />
regarding their effectiveness.<br />
Government Legislation <strong>and</strong> Regulation<br />
The concept <strong>of</strong> patient safety has been championed by several prominent legislators in<br />
both major political parties <strong>and</strong> has become the topic <strong>of</strong> a great deal <strong>of</strong> national debate. Proposals<br />
to date include the establishment <strong>of</strong> voluntary <strong>and</strong> m<strong>and</strong>atory error reporting systems, the<br />
publication <strong>of</strong> outcomes data, <strong>and</strong> the development <strong>of</strong> several Health Care Financing<br />
Administration (HCFA) programs to prevent medical errors. 5 In addition, in 2000 the Agency for<br />
Healthcare Research <strong>and</strong> Quality (AHRQ) established a Center for Quality Improvement <strong>and</strong><br />
<strong>Patient</strong> <strong>Safety</strong>, whose m<strong>and</strong>ate in part is to fund research <strong>and</strong> demonstration projects in patient<br />
safety. Though most <strong>of</strong> these Federal efforts are in the formative stages, 6 successful Federal<br />
agency <strong>and</strong> regulatory efforts outside <strong>of</strong> medicine, most notably in workplace safety <strong>and</strong><br />
commercial airline travel, may herald medicine’s success. In these fields, the Occupational<br />
<strong>Safety</strong> <strong>and</strong> Health Administration (OSHA) <strong>and</strong> the Federal Aviation Administration (FAA) are<br />
7, 8<br />
credited with catalyzing significant improvements in safety over the past 30 years.<br />
Despite a paucity <strong>of</strong> data regarding the effect <strong>of</strong> State legislation on medical errors <strong>and</strong><br />
patient safety, there is some evidence regarding the effectiveness <strong>of</strong> State regulatory efforts to<br />
improve health care quality. 9-11 Among the most prominent set <strong>of</strong> regulations are the New York<br />
State enactments as a result <strong>of</strong> the death <strong>of</strong> Libby Zion.<br />
The daughter <strong>of</strong> a prominent reporter, Ms. Zion died soon after being admitted to the<br />
medical service <strong>of</strong> a New York City hospital. A detailed investigation <strong>of</strong> the circumstances <strong>of</strong><br />
her death subsequently raised concerns regarding working conditions <strong>and</strong> the supervision <strong>of</strong><br />
resident physicians. The Bell Commission, formed at the behest <strong>of</strong> the New York State Health<br />
Commissioner, later recommended major reform in these areas <strong>and</strong> several regulations were<br />
enacted m<strong>and</strong>ating dramatic changes in working conditions <strong>and</strong> supervisory oversight. 12<br />
These regulations markedly altered resident physician education in New York State.<br />
Although they anecdotally resulted in an improvement in resident morale <strong>and</strong> quality <strong>of</strong> life, 13<br />
601
their effect on patient safety is less certain. One retrospective cohort study demonstrated that<br />
patients treated after the work-hour limitations were instituted were more likely to suffer from<br />
complications <strong>and</strong> delays in the performance <strong>of</strong> diagnostic tests. 11 A second retrospective<br />
analysis <strong>of</strong> patient transfer-<strong>of</strong>-care, a bi-product <strong>of</strong> restricting resident work-hours, showed<br />
increased lengths <strong>of</strong> stay <strong>and</strong> utilization <strong>of</strong> laboratory testing for patients that were “h<strong>and</strong>ed<br />
<strong>of</strong>f.” 10 A third study, however, revealed exactly contradictory results. It found the work-hours<br />
limitations led to shorter lengths <strong>of</strong> stay <strong>and</strong> fewer medication errors. 9 These studies have been<br />
criticized for concentrating on the work-hour regulations when the main finding <strong>of</strong> the Bell<br />
Commission was that increased supervision <strong>of</strong> resident physicians was a more important<br />
initiative. 14 (See Chapter 46 for a more complete discussion <strong>of</strong> fatigue <strong>and</strong> work hours).<br />
Although regulations to improve patient safety might be a more efficient way <strong>of</strong><br />
changing practice than less coercive methods such as education or feedback, the use <strong>of</strong><br />
government regulation in the assurance <strong>of</strong> patient safety has limitations. A primary concern is<br />
that regulations may be crafted by legislators who lack intimate knowledge <strong>of</strong> the health care<br />
system. In addition, health care in the United States is extremely heterogeneous - what may be<br />
feasible <strong>and</strong> appropriate in one setting may be inapplicable in another. The differences between<br />
the delivery <strong>of</strong> care in urban <strong>and</strong> rural settings may be particularly troublesome in this regard.<br />
Finally, it is unclear whether government agencies would provide adequate funding to assist<br />
health care organizations to comply with new regulations. For example, a cash-starved institution<br />
faced with a resident work-hours m<strong>and</strong>ate might need to decrease nurse staffing or defer<br />
purchase <strong>of</strong> a computerized order entry system to meet the m<strong>and</strong>ate.<br />
Government agencies may also influence patient safety practices through means other<br />
than direct legislation. For example, the National Nosocomial Infections Surveillance (NNIS)<br />
system may serve as a template for the development <strong>of</strong> a government-sponsored safety program.<br />
Established <strong>and</strong> administered by the Centers for Disease Control <strong>and</strong> Prevention (CDC), the<br />
NNIS is a voluntary nationwide consortium <strong>of</strong> 300 hospitals that regularly report the incidence<br />
<strong>of</strong> specified nosocomial infections. 15 Through analysis <strong>of</strong> the aggregate data, benchmarks are set<br />
for the expected rates <strong>of</strong> nosocomial infection that hospitals may then strive to meet or better.<br />
There is some evidence that the NNIS has contributed to a substantial decline in the rate <strong>of</strong><br />
nosocomial infections over the past several decades. 16 It is conceivable that a similar program<br />
could be established for broader patient safety issues. Although voluntary <strong>and</strong> lacking<br />
enforcement power, NNIS-like patient safety benchmarking could significantly improve safety,<br />
especially if the data were made available to the public or accreditation agencies.<br />
Infection control <strong>of</strong>ficers, another aspect <strong>of</strong> the CDC-championed infection control<br />
movement, may also be applicable to the patient safety movement. Presently, infection control<br />
<strong>of</strong>ficers employed by medical centers focus on improving the system <strong>and</strong> changing practices to<br />
decrease the institutional rates <strong>of</strong> serious infections. An institutional “patient safety <strong>of</strong>ficer”<br />
might assume an analogous function with regard to decreasing adverse events <strong>and</strong> errors. The<br />
establishment <strong>of</strong> a patient safety <strong>of</strong>ficer or committee is one <strong>of</strong> the new Joint Commission on<br />
Accreditation <strong>of</strong> Healthcare Organization’s (JCAHO) safety st<strong>and</strong>ards (discussed below). To<br />
date, there are no data regarding the effectiveness <strong>of</strong> these practices in patient safety.<br />
Accreditation Organizations<br />
Accreditation organizations represent another sector <strong>of</strong> the health care industry that is<br />
assuming new responsibilities in the field <strong>of</strong> patient safety. JCAHO, an outgrowth <strong>of</strong><br />
accreditation programs initiated by the American College <strong>of</strong> Surgeons several decades ago, is the<br />
best known <strong>of</strong> these organizations. JCAHO conducts meticulous inspections <strong>of</strong> medical centers,<br />
602
hospitals <strong>and</strong> other health care institutions on a triennial basis. Survey results are subsequently<br />
used in the process <strong>of</strong> obtaining Medicare certification <strong>and</strong> State licensure <strong>and</strong> to obtain bond<br />
ratings <strong>and</strong> managed care contracts. Although such inspections had previously included some<br />
elements relating to the field <strong>of</strong> patient safety (including infection control <strong>and</strong> the prevention <strong>of</strong><br />
medication errors), they tended to focus on organizational topics including information<br />
management, institutional leadership <strong>and</strong> strategic planning. 17 In response to concerns regarding<br />
patient safety, however, JCAHO has recently launched a major patient safety initiative <strong>and</strong><br />
implemented an entirely new set <strong>of</strong> st<strong>and</strong>ards in July 2001. 18<br />
The new JCAHO st<strong>and</strong>ards place a much greater emphasis on the prevention <strong>of</strong> medical<br />
errors <strong>and</strong> the process <strong>of</strong> responding to medical errors once they occur. A particular focus <strong>of</strong> the<br />
new initiative is the development <strong>of</strong> organization-specific patient safety programs. Such<br />
programs, which will undoubtedly require substantial resources to implement, are expected to<br />
have well-defined leadership, to proactively determine areas where errors are likely to occur, <strong>and</strong><br />
to be capable <strong>of</strong> effecting significant organizational change when necessary. The key elements <strong>of</strong><br />
these st<strong>and</strong>ards are listed in Table 55.1.<br />
Although there is no published evidence that the patient safety st<strong>and</strong>ards previously<br />
required by JCAHO have reduced medical errors, it seems reasonable to assume they have had<br />
some salutary effect. 6 Because JCAHO reports are now publicly available <strong>and</strong> are used by a wide<br />
variety <strong>of</strong> credentialing agencies <strong>and</strong> health care purchaser organizations, many hospitals <strong>and</strong><br />
other medical institutions will make serious <strong>and</strong> concerted efforts to meet the new st<strong>and</strong>ards.<br />
Although lacking the enforcement power <strong>of</strong> Federal regulations, the agencies’ knowledge <strong>of</strong>, <strong>and</strong><br />
contacts within the medical community may produce change more efficiently. Moreover, the<br />
process <strong>of</strong> accreditation involves frequent site visits between the agencies <strong>and</strong> health care<br />
organizations. These visits, in turn, may allow for interactions between institutions <strong>and</strong><br />
accreditors which, under ideal circumstances, could allow for user input in to modifications <strong>of</strong><br />
regulations. How <strong>of</strong>ten this ideal is realized is not known, <strong>and</strong> some observers have questioned<br />
JCAHO’s overall effectiveness. 19-21<br />
Health Care Purchaser Initiatives<br />
The business community has also reacted to the perceived crisis <strong>of</strong> safety in health care.<br />
The most prominent example is the Leapfrog Group. 3 Sponsored by a consortium <strong>of</strong> major<br />
corporate chief executive <strong>of</strong>ficers known as the Business Roundtable, the Leapfrog Group’s<br />
stated commitment is “to mobilize employer purchasing power to initiate breakthrough<br />
improvements in the safety <strong>of</strong> health care for Americans.” 22 Large volume purchasers <strong>of</strong> health<br />
care, including Aetna, ATT, IBM <strong>and</strong> many other Fortune 500 companies, have joined the group.<br />
Combined, their annual health care outlay is over $45 billion. Using their considerable financial<br />
influence, the group hopes to impact medical care by requiring or incentivizing health care<br />
providers to adhere to certain st<strong>and</strong>ards for the process <strong>and</strong> delivery <strong>of</strong> care. 22<br />
Thus far the Leapfrog consortium has chosen to promote 3 patient safety practices: the<br />
use <strong>of</strong> computerized physician order entry (Chapter 6), the involvement <strong>of</strong> critical care<br />
physicians in the care <strong>of</strong> intensive care unit patients (Chapter 38), <strong>and</strong> the use <strong>of</strong> evidence-based<br />
hospital referral systems (Chapter 18). 23 The latter practice refers to the referral <strong>of</strong> patients to<br />
hospitals with the highest volume <strong>and</strong> best outcome figures for certain elective medical<br />
procedures <strong>and</strong> treatments. These initiatives were selected because there is substantial evidence<br />
that they enhance quality <strong>of</strong> care <strong>and</strong> their implementation is both feasible <strong>and</strong> easily assessed.<br />
Initial research sponsored by the group suggests that implementing just these 3 strategies could<br />
prevent almost 60,000 deaths per year <strong>and</strong> avoid over 500,000 errors in medication<br />
603
administration. 24, 25 It is anticipated that other quality-related practices (some directed at patient<br />
safety targets) will be added to the list as evidence for their effectiveness is accumulated.<br />
The Leapfrog Group has also begun to outline a program to improve compliance with the<br />
target practices. Plans include rewarding complying providers with an increased number <strong>of</strong><br />
patients, increasing remuneration for specific services, <strong>and</strong> providing public recognition. General<br />
quality will be assessed <strong>and</strong> publicized through the use <strong>of</strong> rankings, including those assigned by<br />
JCAHO <strong>and</strong> other accreditation organizations. 23 In addition, the employees <strong>of</strong> participating<br />
companies will be encouraged to become more active in choosing providers that meet the<br />
Leapfrog st<strong>and</strong>ards.<br />
Although the Leapfrog initiative is in its nascent stages <strong>and</strong> there is presently no<br />
objective evidence that it will favorably impact patient safety, the sheer financial clout <strong>of</strong> the<br />
involved corporations may catalyze rapid <strong>and</strong> meaningful change. As with government<br />
regulation, it is unclear which sector <strong>of</strong> the health care industry will bear the brunt <strong>of</strong> the<br />
implementation costs. The institution <strong>of</strong> computerized order entry systems, for example, will<br />
require substantial financial outlays (several million dollars in hardware, s<strong>of</strong>tware <strong>and</strong> training<br />
for the average hospital; see Chapter 6) that hospitals <strong>and</strong> medical groups may find extremely<br />
difficult to absorb without assistance. Additionally, physicians <strong>and</strong> other health care providers<br />
may resist changes forced upon them from outside medicine, or may “game the system” to create<br />
the appearance <strong>of</strong> change simply to meet the st<strong>and</strong>ards. 26 Finally, purchaser initiatives may<br />
create disparities in the level <strong>of</strong> patient safety among socioeconomic groups if the changes they<br />
promote are only required <strong>of</strong> health care institutions that provide care to insured members <strong>of</strong> the<br />
group.<br />
Other Approaches<br />
Pr<strong>of</strong>essional Societies<br />
Some <strong>of</strong> the earliest efforts to improve patient safety were actually directed by medical<br />
pr<strong>of</strong>essional societies rather than outside regulators, accreditors, or legislative bodies. The<br />
American Society <strong>of</strong> Anesthesiologists (ASA), for example, formed the Anesthesia <strong>Patient</strong><br />
<strong>Safety</strong> Foundation in 1984 <strong>and</strong> has since promulgated a number <strong>of</strong> reforms that have<br />
substantially changed the routine practice <strong>of</strong> anesthesia. <strong>Safety</strong> measures such as continuous<br />
electrocardiographic monitoring, pulse oximetry <strong>and</strong> preoperative evaluation were strongly<br />
championed by these organizations through the dissemination <strong>of</strong> quality st<strong>and</strong>ards <strong>and</strong><br />
newsletters. 1 While no evidence directly links these initiatives to improved patient safety, there<br />
is little doubt that these reforms resulted in substantial advances. In fact, the st<strong>and</strong>ards <strong>of</strong> care<br />
promoted by the ASA have been widely adopted <strong>and</strong> now represent the recognized minimum<br />
level <strong>of</strong> appropriate care. 4 It is nonetheless difficult to separate the effects <strong>of</strong> these st<strong>and</strong>ards<br />
from secular trends associated with technological <strong>and</strong> clinical advances.<br />
Although medical pr<strong>of</strong>essional societies do not possess the regulatory might <strong>of</strong> the<br />
Federal government or the financial power <strong>of</strong> large health care purchasers, their st<strong>and</strong>ing in the<br />
medical community <strong>and</strong> collective clinical experience are great advantages. It is well established<br />
that physicians are more apt to follow practice guidelines sponsored by respected medical<br />
societies than those issued by the government or industry. 27 Society-based programs are also<br />
likely to allow for more provider input <strong>and</strong> may be more amenable to frequent modification.<br />
Unfortunately, because they lack the power to compel, society recommendations cannot be the<br />
sole agent <strong>of</strong> change, especially when the evidence supporting a practice is extremely strong <strong>and</strong><br />
the stakes are high. It is important to recognize that society recommendations carry a potential<br />
604
for bias, particularly when the recommended changes may have an economic impact on society<br />
members.<br />
<strong>Public</strong>ation <strong>of</strong> Performance Data<br />
Each <strong>of</strong> the large entities described above (legislatures, accrediting bodies, purchasers,<br />
payers, <strong>and</strong> pr<strong>of</strong>essional societies) may choose to disseminate performance data to the public as<br />
part <strong>of</strong> its quality improvement strategy. To date, most such report cards have focused on<br />
discrete quality outcomes for single diseases or procedures (eg, mortality after coronary<br />
angioplasty) rather than patient safety targets such as error or nosocomial infection rates.<br />
Nonetheless, implicit in the vigorous debate regarding m<strong>and</strong>atory error reporting systems is the<br />
question <strong>of</strong> whether public reporting <strong>of</strong> performance data is effective in either motivating<br />
provider change or facilitating informed choices by patients.<br />
Marshall <strong>and</strong> colleagues recently reviewed the evidence regarding the impact <strong>of</strong> public<br />
reporting systems. 28 They found that such systems had a relatively small impact on patients (the<br />
potential users <strong>of</strong> the data), but a greater impact on the hospitals (the sources <strong>of</strong> the data). They<br />
posit that the impact is growing as the public becomes increasingly comfortable with both the<br />
concept <strong>and</strong> interpretation <strong>of</strong> quality reports. Although some have claimed that public reporting<br />
systems, such as New York State’s Cardiac Surgery Data System (which has reported riskadjusted<br />
coronary bypass outcomes since 1990), have led to major improvements in quality, 29<br />
this remains controversial. 30 Proponents point to New York’s falling bypass mortality rate as<br />
evidence <strong>of</strong> the value <strong>of</strong> public reporting, but there is some evidence that the fall in the rate was<br />
due to outmigration <strong>of</strong> high-risk patients to other states rather than true quality improvement. 31<br />
Comment<br />
Concerns about medical errors have spurred many organizations <strong>and</strong> institutions to<br />
launch major patient safety initiatives. Perhaps because they represent a relatively recent<br />
development or because empirical measurement <strong>of</strong> such macro-changes is difficult (isolating the<br />
effects <strong>of</strong> individual interventions vs. other confounding influences), there is little objective<br />
evidence either to determine if they will result in meaningful change or to consider their relative<br />
advantages <strong>and</strong> disadvantages (Table 55.2). Yet Federal <strong>and</strong> State governments, accreditation<br />
agencies such as JCAHO, <strong>and</strong> health care purchasers such as the Leapfrog Coalition, in<br />
combination <strong>and</strong> independently, may eventually be highly effective champions <strong>of</strong> patient safety<br />
initiatives. In addition, pr<strong>of</strong>essional medical societies, given their influential role in the medical<br />
community, are effective agents <strong>of</strong> change in certain circumstances. The work <strong>and</strong> involvement<br />
<strong>of</strong> these diverse, powerful organizations <strong>and</strong> institutions may prove to be valuable adjuncts to the<br />
more traditional mechanisms <strong>of</strong> change represented by practice guidelines, continuing medical<br />
education programs, <strong>and</strong> decision support systems.<br />
605
Table 55.1. New JCAHO safety st<strong>and</strong>ards<br />
-Development <strong>of</strong> a leadership individual or group to devise <strong>and</strong> implement a<br />
comprehensive patient safety program<br />
-Development <strong>of</strong> a proactive error prevention program that includes means <strong>of</strong> identifying<br />
potentially high risk areas<br />
-Development <strong>of</strong> systems for the reporting <strong>of</strong> errors<br />
-Development <strong>of</strong> an error-response system including protocols for root cause analysis<br />
-Requirement for an annual report discussing errors, the response to errors <strong>and</strong> the<br />
programs initiated to prevent future errors<br />
-Requirement for hospital leaders to set “measurable objectives” for patient safety<br />
programs<br />
-Requirement for educational initiatives for employees, stressing the concept <strong>of</strong> patient<br />
safety<br />
606
Table 55.2. A comparison <strong>of</strong> non-local methods to promote patient safety practices<br />
Approach Example Advantages Disadvantages<br />
Legislation “Libby Zion”<br />
laws limiting<br />
resident work<br />
hours<br />
Accreditation JCAHO patient<br />
safety st<strong>and</strong>ards<br />
Market-based The Leapfrog<br />
Group<br />
Pr<strong>of</strong>essional<br />
Societies<br />
Anesthesia<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Foundation<br />
-potential for widespread<br />
implementation<br />
-supported by government<br />
enforcement ability<br />
-may be more flexible <strong>and</strong><br />
more easily modified than<br />
legislation<br />
-implemented at the level <strong>of</strong><br />
the health care organization<br />
-health care providers may<br />
have the opportunity for input<br />
-uses the power <strong>of</strong> the market<br />
to induce change (may be<br />
more acceptable for many<br />
providers than regulatory<br />
solutions)<br />
-may involve carrot (eg, higher<br />
payments for better practices<br />
or outcomes) rather than stick<br />
alone to achieve impact<br />
-readily accepted by health<br />
care providers<br />
-developed by providers<br />
themselves, leading to better<br />
“buy in”<br />
-more easily modified when<br />
new evidence or changes in<br />
practice emerge<br />
607<br />
-inflexible<br />
-limited acceptance by health<br />
care providers<br />
-potential to be developed with<br />
inadequate input from providers<br />
<strong>and</strong> experts<br />
-may be politically driven with<br />
limited applicability<br />
-may not provide for costs <strong>of</strong><br />
implementation, leading to costshifting<br />
away from other<br />
beneficial patient safety<br />
practices<br />
-dependent on voluntary<br />
participation in the accreditation<br />
process<br />
-limited enforcement ability<br />
-generally assessed only every<br />
few years<br />
-potential to cause disparity in<br />
care among groups not covered<br />
by initiatives<br />
-limited acceptance by health<br />
care providers<br />
-potential for st<strong>and</strong>ards to<br />
develop with inadequate input<br />
from health care providers<br />
-change is not required, <strong>and</strong><br />
therefore implementation may<br />
be limited<br />
-minimal enforcement potential;<br />
depends largely on voluntary<br />
participation by practitioners<br />
-potential for bias by<br />
pr<strong>of</strong>essional societies
References<br />
1. Kohn LT, Corrigan JM, Donaldson MS. To Err is Human: Building a Safer Health System.<br />
Washington, DC; Committee on Quality <strong>of</strong> Health Care in America: Institute <strong>of</strong> Medicine;<br />
2000<br />
2. Jencks SF. The government's role in hospital accountability for quality <strong>of</strong> care. Jt Comm J<br />
Qual Improv. 1994; 20: 364-9.<br />
3. Milstein A, Galvin RS, Delbanco SF, Salber P, Buck CR. Improving the safety <strong>of</strong> health<br />
care: the leapfrog initiative. Eff Clin Pract. 2000; 3: 313-6.<br />
4. Gaba DM. Anaesthesiology as a model for patient safety in health care. BMJ. 2000; 320:<br />
785-8.<br />
5. Parsons DW. Federal legislation efforts to improve patient safety. Eff Clin Pract. 2000; 3:<br />
309-12.<br />
6. Brennan TA. The role <strong>of</strong> regulation in quality improvement. Milbank Q. 1998; 76: 709-31,<br />
512.<br />
7. Occupational <strong>Safety</strong> <strong>and</strong> Health Administration; About OSHA. Available at:<br />
http://www.osha.gov/as/opa/osha30yearmilestones.html. Accessed May 8, 2001.<br />
8. NTSB Aviation Accident/Incident Database: Accidents <strong>and</strong> Accident Rates by NTSB<br />
Classification Federal Aviation Administration. Available at:<br />
http://nasdac.faa.gov/asp/fw_ntsb.asp. Accessed May 8, 2001.<br />
9. Gottlieb DJ, Parenti CM, Peterson CA, L<strong>of</strong>gren RP. Effect <strong>of</strong> a change in house staff work<br />
schedule on resource utilization <strong>and</strong> patient care. Arch Intern Med. 1991; 151: 2065-70.<br />
10. L<strong>of</strong>gren RP, Gottlieb D, Williams RA, Rich EC. Post-call transfer <strong>of</strong> resident<br />
responsibility: its effect on patient care. J Gen Intern Med. 1990; 5: 501-5.<br />
11. Laine C, Goldman L, Soukup JR, Hayes JG. The impact <strong>of</strong> a regulation restricting medical<br />
house staff working hours on the quality <strong>of</strong> patient care. JAMA. 1993; 269: 374-8.<br />
12. DeBuono BA, Osten WM. The medical resident workload: the case <strong>of</strong> New York State.<br />
JAMA. 1998; 280: 1882-3.<br />
13. Piver CJ. Residency training under New York's new regulations. N Engl J Med. 1991; 324:<br />
273-4.<br />
14. Bell BM. Supervision, not regulation <strong>of</strong> hours, is the key to improving the quality <strong>of</strong><br />
patient care. JAMA. 1993; 269: 403-4.<br />
15. National Nosocomial Infections Surveillance (NNIS) System report, data summary from<br />
January 1990-May 1999, issued June 1999. Am J Infect Control. 1999; 27: 520-32.<br />
16. Monitoring hospital-acquired infections to promote patient safety–United States, 1990-<br />
1999. MMWR Morb Mortal Wkly Rep. 2000; 49: 149-53.<br />
17. Braun BI, Koss RG, Loeb JM. Integrating performance measure data into the Joint<br />
Commission accreditation process. Eval Health Pr<strong>of</strong>. 1999; 22: 283-97.<br />
18. Joint Commission on Accreditation <strong>of</strong> Healthcare Organizations, St<strong>and</strong>ards, St<strong>and</strong>ards<br />
Revisions for 2001, Leadership Chapter. Available at:<br />
http://www.jcaho.org/st<strong>and</strong>ards_frm.html. Accessed May 3, 2001.<br />
19. Weiner CL. The Elusive Quest: Accountability in Hospitals. New York: Aldine De<br />
Gruyter; 2000.<br />
20. Dame L, Wolfe SM. The failure <strong>of</strong> "private" hospital regulation: An analysis <strong>of</strong> the Joint<br />
Commission on Accreditation <strong>of</strong> Healthcare Organizations' inadequate oversight <strong>of</strong><br />
hospitals. Washington DC: <strong>Public</strong> Citizen Health Research Group; 1996.<br />
608
21. Yessian M. The external review <strong>of</strong> hospital quality: the role <strong>of</strong> accreditation (OEI-01-97-<br />
00051). Boston: Office <strong>of</strong> Evaluations <strong>and</strong> Inspections, Office <strong>of</strong> Inspector General,<br />
Department <strong>of</strong> Health <strong>and</strong> Human Services; 1999.<br />
22. The Leapfrog Group for <strong>Patient</strong> <strong>Safety</strong>: About Us. Available at:<br />
http://www.leapfroggroup.org/about.htm. Accessed May 3, 2001.<br />
23. The Leapfrog Group. The Leapfrog Group for <strong>Patient</strong> <strong>Safety</strong>: <strong>Patient</strong> <strong>Safety</strong>. Available at:<br />
http://www.leapfroggroup.org/safety1.htm. Accessed May 3, 2001.<br />
24. Young MP, Birkmeyer JD. Potential reduction in mortality rates using an intensivist model<br />
to manage intensive care units. Eff Clin Pract. 2000; 3: 284-9.<br />
25. Birkmeyer JD B, CM, Wennberg DE, Young M. The Leapfrog Group. Leapfrog <strong>Patient</strong><br />
<strong>Safety</strong> St<strong>and</strong>ards: The Potential Benefit <strong>of</strong> Univeral Adoption. Available at:<br />
http://www.leapfroggroup.org/PressEvent/Birkmeyer_ExecSum.PDF. Accessed May 3,<br />
2001.<br />
26. Simborg DW. DRG creep: a new hospital-acquired disease. N Engl J Med. 1981; 304:<br />
1602-4.<br />
27. Hayward RS, Guyatt GH, Moore KA, McKibbon KA, Carter AO. Canadian physicians'<br />
attitudes about <strong>and</strong> preferences regarding clinical practice guidelines. CMAJ. 1997; 156:<br />
1715-23.<br />
28. Marshall MN, Shekelle PG, Leatherman S, Brook RH. The public release <strong>of</strong> performance<br />
data: what do we expect to gain? A review <strong>of</strong> the evidence. JAMA. 2000; 283: 1866-74.<br />
29. Mukamel DB, Mushlin AI. Quality <strong>of</strong> care information makes a difference: an analysis <strong>of</strong><br />
market share <strong>and</strong> price changes after publication <strong>of</strong> the New York State Cardiac Surgery<br />
Mortality Reports. Med Care. 1998; 36: 945-54.<br />
30. Hannan TJ. Detecting adverse drug reactions to improve patient outcomes. Int J Med Inf.<br />
1999; 55: 61-4.<br />
31. Omoigui NA, Miller DP, Brown KJ, Annan K, Cosgrove D, 3rd, Lytle B, et al.<br />
Outmigration for coronary bypass surgery in an era <strong>of</strong> public dissemination <strong>of</strong> clinical<br />
outcomes. Circulation. 1996; 93: 27-33.<br />
609
610
PART V. ANALYZING THE PRACTICES<br />
Chapter 56. Methodology for Summarizing the Evidence for the <strong>Practices</strong><br />
Chapter 57. <strong>Practices</strong> Rated by Strength <strong>of</strong> Evidence<br />
Chapter 58. <strong>Practices</strong> Rated by Research Priority<br />
Chapter 59. Listing <strong>of</strong> All <strong>Practices</strong>, Categorical Ratings, <strong>and</strong> Comments<br />
611
612
Chapter 56. Methodology for Summarizing the Evidence for the <strong>Practices</strong><br />
Background<br />
The Agency for Healthcare Research <strong>and</strong> Quality (AHRQ) charged the UCSF-Stanford<br />
Evidence-based Practice Center with the task <strong>of</strong> rating or grading the patient safety practices<br />
identified <strong>and</strong> evaluated in this Report. The Report is an anthology <strong>of</strong> diverse <strong>and</strong> extensive<br />
patient safety practices, grouped by general topic (Part III, Sections A-H), <strong>and</strong> then further sorted<br />
within chapters by individual practices. Synthesis was challenging, but critical in order that<br />
readers <strong>and</strong> health care decision-makers could make judgments about practices to implement<br />
<strong>and</strong>/or research further. Keeping these two audiences in mind, we set 3 goals for “rating” the<br />
practices, as follows:<br />
• Develop a framework for rating the main elements <strong>of</strong> practices within the<br />
constraints <strong>of</strong> the available literature, providing as much pragmatic<br />
information as possible to decision-makers who might endorse practices or<br />
fund further research in the area <strong>of</strong> patient safety interventions;<br />
• Document the limitations <strong>of</strong> the rating method so that those “taking home”<br />
messages from the report underst<strong>and</strong> the inherent limitations <strong>of</strong> making<br />
comparisons <strong>of</strong> a highly heterogeneous field <strong>of</strong> possible practices; <strong>and</strong><br />
• Convey the results <strong>of</strong> the ratings in an organized, visually appealing,<br />
accessible way that ensures that our cautionary notes regarding<br />
oversimplifying the ratings are clear.<br />
Ultimately, we aimed to weight the practices, based on the evidence, on a range <strong>of</strong><br />
dimensions, without implying any ability to calibrate a finely gradated scale for those practices<br />
in between. Proper metrics for these comparisons (eg, cost-effectiveness analysis) require more<br />
data than are currently available in the literature.<br />
Data Inputs into the Practice Ratings<br />
For each practice, information about various inputs into the final “roll-up”, as we referred<br />
to our scoring <strong>of</strong> the practices, were prospectively determined. The decision about what<br />
information to attempt to gather was based on the potential expected uses <strong>of</strong> summary tables <strong>of</strong><br />
practices. Three major categories <strong>of</strong> information were gathered to inform the rating exercise:<br />
• Potential Impact <strong>of</strong> the Practice: based on prevalence <strong>and</strong> severity <strong>of</strong> the<br />
patient safety target, <strong>and</strong> current utilization <strong>of</strong> the practice<br />
• Strength <strong>of</strong> the Evidence Supporting the Practice: including an assessment <strong>of</strong><br />
the relative weight <strong>of</strong> the evidence, effect size, <strong>and</strong> need for vigilance to<br />
reduce any potential negative collateral effects <strong>of</strong> practice implementation<br />
• Implementation: considering costs, logistical barriers, <strong>and</strong> policy issues<br />
Further clarification <strong>of</strong> the 3 categories is in order. Authors were asked to report on<br />
prevalence <strong>and</strong> severity <strong>of</strong> the safety target for a given practice in order to categorize the<br />
potential impact <strong>of</strong> implementing the practice. We added to this an assessment <strong>of</strong> the practice’s<br />
potential impact by reviewing evidence <strong>of</strong> its current utilization. If an intervention is already<br />
widely used, the room for improvement, stated in terms <strong>of</strong> additional reductions in adverse<br />
events targeted by the practice that could be achieved by wider implementation, is less than if<br />
few are currently using the practice. Thus, potential impact <strong>of</strong> implementing the practice is a<br />
613
function <strong>of</strong> the prevalence <strong>and</strong> severity <strong>of</strong> the patient safety target (eg, medical error) <strong>and</strong> the<br />
current utilization <strong>of</strong> the practice.<br />
Of course the actual impact <strong>of</strong> any practice is assessable only if factors related to<br />
evidence supporting the practice are evaluated. Since the Report represents an assemblage <strong>of</strong><br />
the evidence for patient safety practices, the instructions to authors outlined the detailed data<br />
elements related to study design <strong>and</strong> outcomes we required them to abstract from the relevant<br />
studies (see Chapter 3). This information was used to assess, in general terms, the overall<br />
strength <strong>of</strong> the studies for each practice. Effectiveness is commonly defined as the net positive<br />
effect in routine practice. Frequently the data reported in the studies related to efficacy, usually<br />
the net positive effect under controlled, experimental situations. The translation from efficacy to<br />
effectiveness is not straightforward if no direct evidence is available, <strong>and</strong> is therefore based on<br />
judgments about the generalizibility <strong>of</strong> the specific research studies conducted. Also <strong>of</strong> key<br />
importance, <strong>and</strong> therefore abstracted from studies for use in the ratings, was the effect size <strong>of</strong> the<br />
intervention.<br />
Finally, evidence-based reviews consider the potential for harm from a medical<br />
intervention, <strong>and</strong> authors were asked to report on any relevant evidence, as well as reasoned<br />
concerns, gleaned from the literature or from common knowledge about a practice.<br />
To address the real-world environment <strong>and</strong> the desire by the public for action in the area<br />
<strong>of</strong> patient safety, practice chapters were designed to include information about cost <strong>and</strong> other<br />
potential barriers to implementation. While authors sometimes discussed cost savings or<br />
reported cost-effectiveness analyses, the focus was on the start-up costs <strong>and</strong> annual outlays for<br />
ongoing use <strong>of</strong> the practice. Although initial <strong>and</strong> ongoing costs are a function <strong>of</strong> local<br />
environments (eg, size <strong>of</strong> the health care network or institution), possible cost savings are likely<br />
to be even more subject to local conditions (eg, prevalence <strong>of</strong> the patient safety target). For<br />
major investment decisions, an assessment <strong>of</strong> trade<strong>of</strong>fs is more appropriate at the local level.<br />
Our intention was simply to report “ballpark” estimates <strong>of</strong> initial <strong>and</strong> recurring costs. Separate<br />
from economic consequence <strong>of</strong> a particular practice implementation are the political <strong>and</strong><br />
technical considerations.<br />
For all <strong>of</strong> these data inputs into the practice ratings, the primary goal was to find the best<br />
available evidence from publications <strong>and</strong> other sources. Because the literature has not been<br />
previously organized with concurrent considerations <strong>of</strong> each <strong>of</strong> these areas, most estimates could<br />
be improved with further research <strong>and</strong> some are informed by only general <strong>and</strong> somewhat<br />
speculative knowledge. Where possible, in the summaries <strong>of</strong> these elements, we have attempted<br />
to highlight assessments made on the basis <strong>of</strong> limited data.<br />
Rating Process<br />
The 4-person Editorial Team developed a rating form that captured the patient safety<br />
target, practice description, <strong>and</strong> general rating categories (eg, High, Medium, Low) for some <strong>of</strong><br />
the elements described in the section above. General heuristics were specified for each category,<br />
although individual judgment for ratings was designed into the process. The form also specified<br />
comment areas to allow raters to document their specific judgments <strong>and</strong> concerns about ratings.<br />
Each chapter was independently rated by each Editor as to the practices for which there was<br />
evidence. The Editorial Team convened for 3 days to compare scores, discuss disparities, <strong>and</strong><br />
come to consensus about ratings–both by category <strong>and</strong> summary ratings–or the reviewed<br />
practices.<br />
614
Details about Decision Rules <strong>and</strong> Judgment Considerations<br />
Potential Impact Factor<br />
As noted above, an assessment <strong>of</strong> potential impact considered the prevalence <strong>and</strong> severity<br />
<strong>of</strong> the patient safety target, <strong>and</strong> the current utilization <strong>of</strong> the practice being evaluated. The<br />
Editorial Team used the data from the chapters <strong>and</strong> clinical knowledge to order the potential<br />
impact as “High,” “Medium,” “Low,” or “Insufficient Information.” To qualify for the “High”<br />
score, a practice had to target a patient population <strong>of</strong> greater than 1% <strong>of</strong> hospitalized patients<br />
(about 300,000 patients per year) or target a patient safety problem that can result in death or<br />
disability. The “Low” score was used for target populations <strong>of</strong> less than 0.01% <strong>of</strong> hospitalized<br />
patients (about 3000 patient/year) who might experience reversible adverse effects if an effective<br />
practice were not available. Potential impact was deemed a “Medium” if the practice had a<br />
patient safety target that fell between the 2 other categories.<br />
An additional decision rule was applied to the Impact rating after the initial assessment<br />
based on prevalence <strong>and</strong> severity was made. If a practice was currently widely used (>75% <strong>of</strong><br />
hospitals), then the rating was demoted one notch (ie, from High to Medium or Medium to Low).<br />
When this situation occurred, a notation identified that the potential impact level was impacted<br />
by its high current utilization.<br />
We reserved the “Insufficient Information” category for those cases where the prevalence<br />
<strong>and</strong> severity information was quite limited or where the patient safety target was ill-defined.<br />
Evidence Supporting the Practice<br />
Study strength, effect size on target(s), <strong>and</strong> need for vigilance due to potential harms<br />
were rated based more on judgment than pre-specified decision rules. In each case, raters<br />
documented their reasons for category choices.<br />
For study strength, the level <strong>of</strong> study design <strong>and</strong> outcomes (see Chapter 3 for<br />
hierarchies), number <strong>of</strong> studies, numbers <strong>of</strong> patients in studies, generalizability, <strong>and</strong> other<br />
methodologic issues were specified as factors to consider in weighting the relative study strength<br />
for a particular practice. Study strength could be categorized as “High,” “Medium,” or “Low.”<br />
The actual findings <strong>of</strong> the studies were not considered when scoring study strength because this<br />
information was captured in the assessment <strong>of</strong> effect size on target. If there was minimal or no<br />
evidence about a practice, the study strength rating was “Low” <strong>and</strong> raters did not score the<br />
remaining 2 elements <strong>of</strong> the evidence supporting the practice since that might give undue<br />
“credit” to the findings.<br />
The assessment <strong>of</strong> effect size on target(s) was based on the relative risk reductions or<br />
odds ratios reported in the reviewed studies for evidence <strong>of</strong> effectiveness. The raters only used<br />
the findings reported in the practice chapters, <strong>and</strong> did not perform additional analyses (eg, metaanalysis).<br />
If all studies or, in cases where there were a large number <strong>of</strong> studies, the vast majority<br />
showed a positive <strong>and</strong> appreciable effect size (ie, greater than 15% relative risk reduction), then<br />
the positive effect size was categorized as “Robust.” If there was clearly no effect or a very<br />
minimal effect (ie, less than 5% relative risk reduction), then the positive effect size was rated as<br />
“Negligible.” For findings that were considered suggestive <strong>of</strong> substantive effect, but not clearly<br />
“Robust,” the category used was “Modest.” The final category, “Unclear,” captured those<br />
practices for which the effect size results were inconsistent.<br />
For any given practice that reduces one adverse event, it is conceivable that new<br />
problems might ensue when the practice is implemented. Thus, we subjectively rated the<br />
615
concern for harm based on the level <strong>of</strong> vigilance necessary to ensure that the practice, if<br />
implemented, would not result in collateral negative effects. The categories available were<br />
“Low,” “Medium,” <strong>and</strong> “High.” Thus, a practice rated as “Low” would require little to no<br />
attentiveness to potential harms, while one rated as “High” would merit heightened monitoring<br />
for potential negative effects. These ratings were made conservatively, meaning that when in<br />
doubt, a higher vigilance category was selected.<br />
Implementation<br />
Assuming a 3-year lead time for implementation, patient safety practices were rated for<br />
their costs <strong>and</strong> complexity. Costs were based on initial start-up <strong>and</strong> annual expenditures for full<br />
implementation at an average size hospital or health care organization. Potential cost savings<br />
were not considered for the rating, but were reported in the practice chapters if they were<br />
documented in the literature. If a practice was expected to require expenditures <strong>of</strong> greater than<br />
about $1 million, the rating was “High.” Expenditures <strong>of</strong> approximately $100,000-$1 million<br />
were categorized as “Medium.” Below this level, practices were rated as “Low” in terms <strong>of</strong> cost.<br />
The feasibility <strong>of</strong> implementation was rated by considering potential political (eg, major<br />
shifts in who delivers care) <strong>and</strong> technical (eg, integration <strong>of</strong> legacy <strong>and</strong> newer computer systems)<br />
obstacles. Because relatively few data exist for rating implementation complexity, we used only<br />
2 categories, “Low” <strong>and</strong> “High,” meaning relatively easy <strong>and</strong> relatively difficult. In cases in<br />
which implementation could be accomplished simply with the expenditure <strong>of</strong> dollars, we gave<br />
high cost scores but low feasibility scores.<br />
Overall Rating for Impact/Evidence<br />
In addition, each member <strong>of</strong> the team considered the totality <strong>of</strong> information on potential<br />
impact <strong>and</strong> evidence supporting the practice to score each on a 0 to 10 scale (“Strength <strong>of</strong> the<br />
Evidence”). For these ratings, we took the perspective <strong>of</strong> a leader <strong>of</strong> a large health care<br />
enterprise (eg, a hospital or integrated delivery system) <strong>and</strong> asked the question, “If you wanted to<br />
improve patient safety at your institution over the next 3 years <strong>and</strong> resources were not a<br />
significant consideration, how would you grade this practice?” For this rating, we explicitly did<br />
not consider difficulty or cost <strong>of</strong> implementation in the rating. Rather, the rating simply<br />
reflected the strength <strong>of</strong> the evidence regarding the effectiveness <strong>of</strong> the practice <strong>and</strong> the probable<br />
impact <strong>of</strong> its implementation on reducing adverse events related to health care exposure. If the<br />
patient safety target was rated as “High” impact <strong>and</strong> there was compelling evidence (ie, “High”<br />
relative study strength) that a particular practice could significantly reduce (eg, “Robust” effect<br />
size) the negative consequences (eg, hospital-acquired infections), raters were likely to score the<br />
practice close to 10. If the studies were less convincing, the effect size was less robust, or there<br />
was a need for a “Medium” or “High” degree <strong>of</strong> vigilance because <strong>of</strong> potential harms, then the<br />
rating would be lower.<br />
Overall Rating for Research Priority<br />
Analogously, we also rated the usefulness <strong>of</strong> conducting more research on each practice,<br />
emphasizing whether there appeared to be questions that a research program might have a<br />
reasonable chance <strong>of</strong> addressing successfully (“Research Priority”). Here, our “thought<br />
question” was, “If you were the leader <strong>of</strong> a large agency or foundation committed to improving<br />
patient safety, <strong>and</strong> were considering allocating funds to promote additional research, how would<br />
you grade this practice?” If there was a simple gap in the evidence that could be addressed by a<br />
research study or if the practice was multifaceted <strong>and</strong> implementation could be eased by<br />
616
determining the specific elements that were effective, then the research priority was high. If the<br />
area was one <strong>of</strong> high potential impact (ie, large number <strong>of</strong> patients at risk for morbid or mortal<br />
adverse events) <strong>and</strong> a practice had been inadequately researched, then it also would also receive<br />
a relatively high rating for research need. <strong>Practices</strong> might receive low research scores if they<br />
held little promise (eg, relatively few patients affected by the safety problem addressed by the<br />
practice or a significant body <strong>of</strong> knowledge already demonstrating the practice’s lack <strong>of</strong> utility).<br />
Conversely, a practice that was clearly effective, low cost <strong>and</strong> easy to implement would not<br />
require further research <strong>and</strong> would also receive low research scores.<br />
Caveats to Ratings<br />
For all elements assessed, divergent assessments among the 4 Editor-raters were<br />
infrequent <strong>and</strong> were discussed until consensus was reached. For each final category where<br />
differences in interpretation existed <strong>and</strong> persisted after discussion, the protocol was to document<br />
a comment about these differences (see Chapter 59). Comments were also noted when specific<br />
additional information could clarify concerns about fidelity <strong>of</strong> a specific rating. In a few cases,<br />
categories that had not been specified were created for unusual circumstances <strong>and</strong> again<br />
comments to explain the category were documented.<br />
Rating Tables<br />
Information Captured<br />
Ratings were recorded on a data table, <strong>and</strong> comments were footnoted. Summarizing from<br />
the previous discussion, the various data tables appearing in Chapters 57-59 captured some or all<br />
<strong>of</strong> the following 11 elements <strong>and</strong> rating categories, ordered from strongest to weakest:<br />
1. Chapter number<br />
2. <strong>Patient</strong> safety target(s)<br />
3. <strong>Patient</strong> safety practice description<br />
4. Potential Impact: High, Medium, Low, Insufficient Information<br />
5. Study Strength: High, Medium, Low<br />
6. Effect Size: Robust, Modest, Negligible, Unclear<br />
7. Vigilance: Low, Medium, High<br />
8. Implementation cost: Low, Medium, High<br />
9. Implementation complexity (political, technical): Low, High<br />
10. Overall rating for impact/evidence: 0 to 10 (10 is highest), in 0.5 increments<br />
11. Overall rating for research need: 0 to 10 (10 is highest), in 0.5 increments<br />
Information Reported<br />
Chapter 59 presents the detailed data tables with categorization <strong>of</strong> elements 1-9 above.<br />
Reporting specific scores for the overall ratings would imply a refinement in scoring that was<br />
neither attempted nor advisable given the nature <strong>of</strong> the information available. Where applicable,<br />
caveats to the categorizations are appended in a series <strong>of</strong> endnotes.<br />
In rating both the strength <strong>of</strong> the evidence <strong>and</strong> the research priority, our purpose was not<br />
to report precise 1-10 scores, but to develop general “zones” or practice groupings. As noted<br />
earlier, better methods are available for comparative ratings when the data inputs are available.<br />
The relative paucity <strong>of</strong> the evidence dissuaded us from using a more precise, sophisticated, but<br />
ultimately unfeasible, approach.<br />
617
Chapter 57 summarizes the overall ratings for the “Strength <strong>of</strong> the Evidence" regarding<br />
their impact <strong>and</strong> effectiveness score, <strong>and</strong> subdivides the practices into 5 zones. <strong>Practices</strong> are<br />
listed from highest score to lowest score for each rating zone. The zones are “greatest strength”<br />
(score <strong>of</strong> 8-10), “high strength” (score <strong>of</strong> 6-7.5), “medium strength” (4-5.5), “lower<br />
impact/evidence scored practices” (score <strong>of</strong> 2-3.5), “lowest impact/evidence scored practices”<br />
(score <strong>of</strong> 0-1.5). <strong>Practices</strong> near the bottom <strong>of</strong> one zone may be just as appropriate to list near the<br />
top <strong>of</strong> the adjacent lower zone. Similarly, practices at the top <strong>of</strong> a zone may actually be more<br />
comparable to those in the adjacent higher zone. The cut-<strong>of</strong>fs between zones are somewhat<br />
artificial, but allow a general <strong>and</strong> reasonable synthesis <strong>of</strong> the data on impact <strong>and</strong> evidence<br />
supporting (or negating) the effects <strong>of</strong> the practice. Readers can be confident that practices that<br />
fall in the highest zone do not belong in the lowest zone.<br />
Chapter 58 summarizes the overall ratings for the “Research Priority” score, <strong>and</strong> provides<br />
examples <strong>of</strong> types <strong>of</strong> research that may be helpful. <strong>Practices</strong> are categorized in 3 zones: “Further<br />
Research Likely to be Highly Beneficial” (scores <strong>of</strong> 7 <strong>and</strong> higher), “Further Research Likely to<br />
be Beneficial” (scores <strong>of</strong> 4 to 6.5 inclusive), <strong>and</strong> “Low Priority for Research” (below 4). The<br />
chapter lists practices for each <strong>of</strong> the top two categories <strong>of</strong> research priority.<br />
618
Chapter 57. <strong>Practices</strong> Rated by Strength <strong>of</strong> Evidence<br />
After rating practices on a metric for potential impact, <strong>and</strong> on the strength <strong>of</strong> the<br />
evidence, we grouped them into 5 categories (Tables 57.1-57.5). These categorizations reflect<br />
the current state <strong>of</strong> the evidence. If a practice that addresses a highly prevalent or severe patient<br />
safety target receives a low rating on the impact/evidence scale, it may be because the strength <strong>of</strong><br />
the evidence base is still weak due to lack <strong>of</strong> evaluations. As a result the practice is likely to<br />
show up at a high level on the research priority scale. However, if the practice has been studied<br />
rigorously, <strong>and</strong> there is clear evidence that its effectiveness is negligible, it is rated at the low<br />
ends <strong>of</strong> both the “strength <strong>of</strong> the evidence” (on impact/effectiveness) scale <strong>and</strong> the “research<br />
priority” scale.<br />
For each practice listed in Tables 57.1 through 57.5, a designation for the cost <strong>and</strong><br />
complexity <strong>of</strong> implementation <strong>of</strong> the practice is included. The ratings for implementation are<br />
“Low,” which corresponds to low cost <strong>and</strong> low complexity (eg, political, technical); “Medium,”<br />
which signifies low to medium cost <strong>and</strong> high complexity, or medium to high cost <strong>and</strong> low<br />
complexity; <strong>and</strong> “High,” which reflects medium to high cost <strong>and</strong> high complexity.<br />
Several practices are not included in the tables because they were not rated. This set <strong>of</strong><br />
practices have long histories <strong>of</strong> use outside <strong>of</strong> medicine, but have not yet received enough<br />
evaluations for their potential health care applications:<br />
• Promoting a Culture <strong>of</strong> <strong>Safety</strong> (Chapter 40)<br />
• Use <strong>of</strong> Human Factors Principles in Evaluation <strong>of</strong> <strong>Medical</strong> Devices (Subchapter 41.1)<br />
• Refining Performance <strong>of</strong> <strong>Medical</strong> Device Alarms (eg, balancing sensitivity <strong>and</strong><br />
specificity <strong>of</strong> alarms, ergonomic design) (Subchapter 41.2)<br />
• Fixed Shifts or Forward Shift Rotations (Chapter 46)<br />
• Napping Strategies (Chapter 46)<br />
619
Table 57.1. <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong> with the Greatest Strength <strong>of</strong> Evidence Regarding their<br />
Impact <strong>and</strong> Effectiveness<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Implementation<br />
Cost/Complex<br />
31 Venous thromboembolism (VTE) Appropriate VTE prophylaxis Low<br />
25 Perioperative cardiac events in<br />
patients undergoing noncardiac<br />
surgery<br />
16.1 Central venous catheter-related<br />
bloodstream infections<br />
Use <strong>of</strong> perioperative beta-blockers Low<br />
Use <strong>of</strong> maximum sterile barriers during<br />
catheter insertion<br />
20.1 Surgical site infections Appropriate use <strong>of</strong> antibiotic prophylaxis Low<br />
48 Missed, incomplete or not fully<br />
comprehended informed consent<br />
Asking that patients recall <strong>and</strong> restate what<br />
they have been told during informed<br />
consent<br />
17.2 Ventilator-associated pneumonia Continuous aspiration <strong>of</strong> subglottic<br />
secretions (CASS)<br />
620<br />
Low<br />
Low<br />
Medium<br />
27 Pressure ulcers Use <strong>of</strong> pressure relieving bedding materials Medium<br />
21 Morbidity due to central venous<br />
catheter insertion<br />
9 Adverse events related to chronic<br />
anticoagulation with warfarin<br />
33 Morbidity <strong>and</strong> mortality in postsurgical<br />
<strong>and</strong> critically ill patients<br />
16.2 Central venous catheter-related<br />
bloodstream infections<br />
Use <strong>of</strong> real-time ultrasound guidance during<br />
central line insertion<br />
<strong>Patient</strong> self management using home<br />
monitoring devices<br />
High<br />
High<br />
Various nutritional strategies Medium<br />
Antibiotic-impregnated catheters Low
Table 57.2 <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong> with High Strength <strong>of</strong> Evidence Regarding their Impact <strong>and</strong><br />
Effectiveness<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Implementation<br />
Cost/Complex<br />
18 Mortality associated with surgical<br />
procedures<br />
Localizing specific surgeries <strong>and</strong> procedures<br />
to high volume centers<br />
17.1 Ventilator-associated pneumonia Semi-recumbent positioning Low<br />
26.5 Falls <strong>and</strong> fall injuries Use <strong>of</strong> hip protectors Low<br />
8 Adverse drug events (ADEs) related<br />
to targeted classes (analgesics, KCl,<br />
antibiotics, heparin) (focus on<br />
detection)<br />
Use <strong>of</strong> computer monitoring for potential<br />
ADEs<br />
20.3 Surgical site infections Use <strong>of</strong> supplemental perioperative oxygen Low<br />
621<br />
High (varies)<br />
Medium<br />
39 Morbidity <strong>and</strong> mortality Changes in nursing staffing Medium<br />
48 Missed or incomplete or not fully<br />
comprehended informed consent<br />
Use <strong>of</strong> video or audio stimuli Low<br />
17.3 Ventilator-associated pneumonia Selective decontamination <strong>of</strong> digestive tract Low<br />
38 Morbidity <strong>and</strong> mortality in ICU<br />
patients<br />
42.1 Adverse events related to<br />
discontinuities in care<br />
15.1 Hospital-acquired urinary tract<br />
infection<br />
Change in ICU structure—active<br />
management by intensivist<br />
Information transfer between inpatient <strong>and</strong><br />
outpatient pharmacy<br />
High<br />
Medium<br />
Use <strong>of</strong> silver alloy-coated catheters Low<br />
28 Hospital-related delirium Multi-component delirium prevention<br />
program<br />
30 Hospital-acquired complications<br />
(functional decline, mortality)<br />
37.4 Inadequate postoperative pain<br />
management<br />
Medium<br />
Geriatric evaluation <strong>and</strong> management unit High<br />
Non-pharmacologic interventions (eg,<br />
relaxation, distraction)<br />
Low
Table 57.3 <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong> with Medium Strength <strong>of</strong> Evidence Regarding their<br />
Impact <strong>and</strong> Effectiveness<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Implementation<br />
Cost/Complex<br />
6 Medication errors <strong>and</strong> adverse drug Computerized physician order entry (CPOE) High<br />
events (ADEs) primarily related to<br />
ordering process<br />
<strong>and</strong> clinical decision support (CDSS)<br />
42.4 Failures to communicate significant Protocols for notification <strong>of</strong> test results to Low<br />
abnormal results (eg, pap smears) patients<br />
47 Adverse events due to transportation<br />
<strong>of</strong> critically ill patients between health<br />
care facilities<br />
Specialized teams for interhospital transport Medium<br />
7 Medication errors <strong>and</strong> adverse drug<br />
events (ADEs) related to ordering <strong>and</strong><br />
monitoring<br />
Clinical pharmacist consultation services Medium<br />
13 Serious nosocomial infections (eg, Barrier precautions (via gowns & gloves; Medium<br />
vancomycin-resistant enterococcus, C.<br />
difficile)<br />
dedicated equipment; dedicated personnel)<br />
20.4 Surgical site infections Perioperative glucose control Medium<br />
34 Stress-related gastrointestinal bleeding H2 antagonists Low<br />
36 Pneumococcal pneumonia Methods to increase pneumococcal<br />
vaccination rate<br />
Low<br />
37.2 Inadequate pain relief Acute pain service Medium<br />
9 Adverse events related to<br />
Anticoagulation services <strong>and</strong> clinics for Medium<br />
anticoagulation<br />
coumadin<br />
14 Hospital-acquired infections due to<br />
antibiotic-resistant organisms<br />
Limitations placed on antibiotic use Low<br />
15.2 Hospital-acquired urinary tract<br />
infection<br />
Use <strong>of</strong> suprapubic catheters High<br />
32 Contrast-induced renal failure Hydration protocols with acetylcysteine Low<br />
35 Clinically significant misread Education interventions <strong>and</strong> continuous Low<br />
radiographs <strong>and</strong> CT scans by nonradiologists<br />
quality improvement strategies<br />
48 Missed or incomplete or not fully Provision <strong>of</strong> written informed consent Low<br />
comprehended informed consent information<br />
49 Failure to honor patient preferences for Computer-generated reminders to discuss Medium (Varies)<br />
end-<strong>of</strong>-life care<br />
advanced directives<br />
9 Adverse events related to<br />
Protocols for high-risk drugs: nomograms Low<br />
anticoagulation<br />
for heparin<br />
622
17.1 Ventilator-associated pneumonia Continuous oscillation Medium<br />
20.2 Surgical site infections Maintenance <strong>of</strong> perioperative normothermia Low<br />
26.2 Restraint-related injury; Falls Interventions to reduce the use <strong>of</strong> physical<br />
restraints safely<br />
Medium<br />
26.3 Falls Use <strong>of</strong> bed alarms Medium<br />
32 Contrast-induced renal failure Use <strong>of</strong> low osmolar contrast media Medium<br />
623
Table 57.4 <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong> with Lower Impact <strong>and</strong>/or Strength <strong>of</strong> Evidence<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Implementation<br />
Cost/Complex<br />
16.3 Central venous catheter-related<br />
bloodstream infections<br />
16.4 Central venous catheter-related<br />
bloodstream infections<br />
16.4 Central venous catheter-related<br />
bloodstream infections<br />
29 Hospital-acquired complications (eg,<br />
falls, delirium, functional decline,<br />
mortality)<br />
37.1 Inadequate pain relief in patients with<br />
abdominal pain in hospital patients<br />
45 Adverse events due to provider<br />
inexperience or unfamiliarity with<br />
certain procedures <strong>and</strong> situations<br />
11 Adverse drug events (ADEs) in drug<br />
dispensing <strong>and</strong>/or administration<br />
Cleaning site (povidone-iodine to<br />
chlorhexidine)<br />
624<br />
Low<br />
Use <strong>of</strong> heparin Low<br />
Tunneling short-term central venous<br />
catheters<br />
Medium<br />
Geriatric consultation services High<br />
Use <strong>of</strong> analgesics in the patient with acute<br />
abdomen without compromising diagnostic<br />
accuracy<br />
Low<br />
Simulator-based training Medium<br />
Use <strong>of</strong> automated medication dispensing<br />
devices<br />
12 Hospital-acquired infections Improve h<strong>and</strong>washing compliance (via<br />
education/behavior change; sink technology<br />
<strong>and</strong> placement; washing substance)<br />
49 Failure to honor patient preferences for<br />
end-<strong>of</strong>-life care<br />
43.1 Adverse events due to patient<br />
misidentification<br />
10 Adverse drug events (ADEs) in<br />
dispensing medications<br />
Use <strong>of</strong> physician order form for lifesustaining<br />
treatment (POLST)<br />
Medium<br />
Low<br />
Low<br />
Use <strong>of</strong> bar coding Medium (Varies)<br />
Unit-dosing distribution system Low<br />
24 <strong>Critical</strong> events in anesthesia Intraoperative monitoring <strong>of</strong> vital signs <strong>and</strong><br />
oxygenation<br />
42.2 Adverse events during cross-coverage St<strong>and</strong>ardized, structured sign-outs for<br />
physicians<br />
44 Adverse events related to team<br />
performance issues<br />
46 Adverse events related to fatigue in<br />
health care workers<br />
Applications <strong>of</strong> aviation-style crew resource<br />
management (eg, Anesthesia Crisis<br />
Management; MedTeams)<br />
Limiting individual provider’s hours <strong>of</strong><br />
service<br />
Low<br />
Low<br />
High<br />
High
57.5 <strong>Patient</strong> <strong>Safety</strong> <strong>Practices</strong> with Lowest Impact <strong>and</strong>/or Strength <strong>of</strong> Evidence<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Implementation<br />
Cost/Complex<br />
23 Complications due to anesthesia<br />
equipment failures<br />
42.3 Adverse events related to information<br />
loss at discharge<br />
Use <strong>of</strong> pre-anesthesia checklists Low<br />
Use <strong>of</strong> structured discharge summaries Low<br />
22 Surgical items left inside patients Counting sharps, instruments <strong>and</strong> sponges Low<br />
17.4 Ventilator-associated pneumonia Use <strong>of</strong> sucralfate Low<br />
26.4 Falls <strong>and</strong> fall-related injuries Use <strong>of</strong> special flooring material in patient<br />
care areas<br />
43.2 Performance <strong>of</strong> invasive diagnostic or<br />
therapeutic procedure on wrong body<br />
part<br />
625<br />
Medium<br />
“Sign your site” protocols Medium<br />
26.1 Falls Use <strong>of</strong> identification bracelets Low<br />
32 Contrast-induced renal failure Hydration protocols with theophylline Low<br />
47 Adverse events due to transportation<br />
<strong>of</strong> critically ill patients within a<br />
hospital<br />
16.4 Central venous catheter-related<br />
bloodstream infections<br />
16.4 Central venous catheter-related<br />
bloodstream infections<br />
Mechanical rather than manual ventilation<br />
during transport<br />
Low<br />
Changing catheters routinely High<br />
Routine antibiotic prophylaxis Medium
626
Chapter 58. <strong>Practices</strong> Rated by Research Priority<br />
Further research on a number <strong>of</strong> practices would clarify a range <strong>of</strong> questions (eg, whether<br />
the practice is effective, what aspects <strong>of</strong> a multi-faceted intervention matter the most, how best to<br />
implement the practice). The conceptual framework for this categorization is described in<br />
Chapter 56. In Table 58.1 <strong>and</strong> 58.2, the practices are grouped in zones: “research likely to be<br />
highly beneficial,” <strong>and</strong> “research likely to be beneficial.” We also list, in the far-right column,<br />
the practices’ categorization for “Strength <strong>of</strong> the Evidence” (as detailed above in Tables 57.1-<br />
57.5). For presentation in this table, this category is simplified into a 1 (“highest strength <strong>of</strong><br />
evidence”) to 5 (“lowest strength <strong>of</strong> evidence”) which corresponds exactly to the groupings in<br />
Tables 57.1-5. We list these here to allow the reader to compare <strong>and</strong> contrast the research<br />
priority rankings with the evidence rankings. <strong>Practices</strong> that are not listed in either Table 58.1 or<br />
58.2 may benefit from more research, but were not scored as highly as those included in these 2<br />
lists.<br />
Table 58.1 Further Research Likely to be Highly Beneficial<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Strength <strong>of</strong><br />
the Evidence<br />
(1-5 Scale; 1<br />
is highest)<br />
20.4 Surgical site infections Perioperative glucose control 3<br />
18 Mortality associated with surgical Localizing specific surgeries <strong>and</strong><br />
2<br />
procedures<br />
procedures to high volume centers<br />
20.3 Surgical site infections Use <strong>of</strong> supplemental perioperative oxygen 2<br />
39 Morbidity <strong>and</strong> mortality Changes in nursing staffing 2<br />
15.1 Hospital-acquired urinary tract<br />
infection<br />
6 Medication errors <strong>and</strong> adverse<br />
drug events (ADEs) primarily<br />
related to ordering process<br />
14 Hospital-acquired infections due<br />
to antibiotic-resistant organisms<br />
Use <strong>of</strong> silver alloy-coated catheters 2<br />
Computerized physician order entry<br />
(CPOE) with clinical decision support<br />
(CDSS)<br />
Limitations placed on antibiotic use 3<br />
20.1 Surgical site infections Appropriate use <strong>of</strong> antibiotic prophylaxis 1<br />
31 Venous thromboembolism (VTE) Appropriate VTE prophylaxis 1<br />
33 Morbidity <strong>and</strong> mortality in postsurgical<br />
<strong>and</strong> critically ill patients<br />
37.1 Inadequate pain relief in patients<br />
with abdominal pain in hospital<br />
patients<br />
Various nutritional strategies (especially<br />
early enteral nutrition in critically ill <strong>and</strong><br />
post-surgical patients)<br />
Use <strong>of</strong> analgesics in the patient with acute<br />
abdomen without compromising<br />
diagnostic accuracy<br />
627<br />
3<br />
1<br />
4
12 Hospital-acquired infections Improve h<strong>and</strong>washing compliance (via<br />
education/behavior change; sink<br />
technology <strong>and</strong> placement; washing<br />
substance)<br />
9 Adverse events related to chronic<br />
anticoagulation with warfarin<br />
21 Morbidity due to central venous<br />
catheter insertion<br />
38 Morbidity <strong>and</strong> mortality in ICU<br />
patients<br />
<strong>Patient</strong> self-management using home<br />
monitoring devices<br />
Use <strong>of</strong> real-time ultrasound guidance<br />
during central line insertion<br />
Change in ICU structure—active<br />
management by intensivist<br />
32 Contrast-induced renal failure Hydration protocols with acetylcysteine 3<br />
43.1 Adverse events due to patient<br />
misidentification<br />
Use <strong>of</strong> bar coding 4<br />
27 Pressure ulcers Use <strong>of</strong> pressure relieving bedding<br />
materials<br />
20.2 Surgical site infections Maintenance <strong>of</strong> perioperative<br />
normothermia<br />
25 Perioperative cardiac events in<br />
patients undergoing noncardiac<br />
surgery<br />
48 Missed or incomplete or not fully<br />
comprehended informed consent<br />
Use <strong>of</strong> perioperative beta-blockers 1<br />
Use <strong>of</strong> video or audio stimuli 2<br />
28 Hospital-related delirium Multi-component delirium prevention<br />
program<br />
7 Medication errors <strong>and</strong> adverse<br />
drug events (ADEs) related to<br />
ordering <strong>and</strong> monitoring<br />
13 Serious nosocomial infections (eg,<br />
vancomycin-resistant<br />
enterococcus, C. difficile)<br />
9 Adverse events related to<br />
anticoagulation<br />
48 Missed, incomplete or not fully<br />
comprehended informed consent<br />
49 Failure to honor patient<br />
preferences for end-<strong>of</strong>-life care<br />
Clinical pharmacist consultation services 3<br />
Barrier precautions (via gowns & gloves;<br />
dedicated equipment; dedicated personnel)<br />
Anticoagulation services <strong>and</strong> clinics for<br />
coumadin<br />
Provision <strong>of</strong> written informed consent<br />
information<br />
Computer-generated reminders to discuss<br />
advanced directives<br />
628<br />
4<br />
1<br />
1<br />
2<br />
1<br />
3<br />
2<br />
3<br />
3<br />
3<br />
3
9 Adverse events related to<br />
anticoagulation<br />
Protocols for high-risk drugs: nomograms<br />
for heparin<br />
26.3 Falls Use <strong>of</strong> bed alarms 3<br />
11 Adverse drug events (ADEs) in<br />
drug dispensing <strong>and</strong>/or<br />
administration<br />
Use <strong>of</strong> automated medication dispensing<br />
devices<br />
629<br />
3<br />
4
Table 58.2 Further Research Likely to be Beneficial<br />
Chapter <strong>Patient</strong> <strong>Safety</strong> Target <strong>Patient</strong> <strong>Safety</strong> Practice Impact/<br />
Evidence<br />
Category<br />
630<br />
(1-5)<br />
17.2 Ventilator-associated Continuous aspiration <strong>of</strong> subglottic 1<br />
pneumonia<br />
secretions (CASS)<br />
17.1 Ventilator-associated<br />
pneumonia<br />
Semi-recumbent positioning 2<br />
26.5 Falls <strong>and</strong> fall injuries Use <strong>of</strong> hip protectors 2<br />
30 Hospital-acquired<br />
Geriatric evaluation <strong>and</strong> management 2<br />
complications (functional<br />
decline, mortality)<br />
unit<br />
47 Adverse events due to Specialized teams for interhospital 3<br />
transportation <strong>of</strong> critically ill<br />
patients between health care<br />
facilities<br />
transport<br />
34 Stress-related gastrointestinal<br />
bleeding<br />
H2-antagonists 3<br />
37.2 Inadequate pain relief Acute pain service 3<br />
15.2 Hospital-acquired urinary tract<br />
infection<br />
Use <strong>of</strong> suprapubic catheters 3<br />
26.2 Restraint-related injury; Falls Interventions to reduce the use <strong>of</strong><br />
physical restraints safely<br />
3<br />
45 Adverse events due to provider<br />
inexperience or unfamiliarity<br />
with certain procedures <strong>and</strong><br />
situations<br />
Simulator-based training 4<br />
49 Failure to honor patient Use <strong>of</strong> physician order form for life- 4<br />
preferences for end-<strong>of</strong>-life care sustaining treatment (POLST)<br />
42.2 Adverse events during cross- St<strong>and</strong>ardized, structured sign-outs for 4<br />
coverage<br />
physicians<br />
44 Adverse events related to team Applications <strong>of</strong> aviation-style crew 4<br />
performance issues<br />
resource management (eg, Anesthesia<br />
Crisis Management; MedTeams)<br />
16.2 Central venous catheter-related<br />
bloodstream infections<br />
17.3 Ventilator-associated<br />
pneumonia<br />
Antibiotic-impregnated catheters 1<br />
Selective decontamination <strong>of</strong> digestive<br />
tract<br />
2
42.4 Failures to communicate<br />
significant abnormal results<br />
(eg, pap smears)<br />
Protocols for notification <strong>of</strong> test results<br />
to patients<br />
36 Pneumococcal pneumonia Methods to increase pneumococcal<br />
vaccination rate<br />
3<br />
16.3 Central venous catheter-related Cleaning site (povidone-iodine to<br />
4<br />
bloodstream infections chlorhexidine)<br />
16.4 Central venous catheter-related<br />
bloodstream infections<br />
Use <strong>of</strong> heparin 4<br />
16.4 Central venous catheter-related Tunneling short-term central venous 4<br />
bloodstream infections catheters<br />
29 Hospital-acquired<br />
complications (eg, falls,<br />
delirium, functional decline,<br />
mortality)<br />
Geriatric consultation services 4<br />
46 Adverse events related to Limiting individual provider’s hours <strong>of</strong> 4<br />
fatigue in health care workers service<br />
26.4 Falls <strong>and</strong> fall-related injuryies Use <strong>of</strong> special flooring material in<br />
patient care areas<br />
5<br />
43.2 Performance <strong>of</strong> invasive<br />
diagnostic or therapeutic<br />
procedure on wrong body part<br />
“Sign your site” protocols 5<br />
42.1 Adverse events related to Information transfer between inpatient 2<br />
discontinuities in care <strong>and</strong> outpatient pharmacy<br />
48 Missed, incomplete or not fully Asking that patients recall <strong>and</strong> restate 1<br />
comprehended informed what they have been told during<br />
consent<br />
informed consent<br />
8 Adverse drug events (ADEs) Use <strong>of</strong> computer monitoring for<br />
2<br />
related to targeted classes<br />
(analgesics, KCl, antibiotics,<br />
heparin) (focus on detection)<br />
potential ADEs<br />
24 <strong>Critical</strong> events in anesthesia Intraoperative monitoring <strong>of</strong> vital signs<br />
<strong>and</strong> oxygenation<br />
4<br />
42.3 Adverse events related to<br />
information loss at discharge<br />
Use <strong>of</strong> structured discharge summaries 5<br />
631<br />
3
632
Chapter 59. Listing <strong>of</strong> All <strong>Practices</strong>, Categorical Ratings, <strong>and</strong><br />
Comments<br />
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
6 Medication errors<br />
<strong>and</strong> adverse drug<br />
events (ADEs)<br />
primarily related<br />
to ordering<br />
process<br />
7 Medication errors<br />
<strong>and</strong> ADEs related<br />
to ordering <strong>and</strong><br />
monitoring<br />
8 ADEs related to<br />
targeted classes<br />
(analgesics, KCl,<br />
antibiotics,<br />
heparin) (focus on<br />
detection)<br />
9 Adverse events<br />
related to<br />
anticoagulation<br />
9 Adverse events<br />
related to<br />
anticoagulation<br />
9 Adverse events<br />
related to chronic<br />
anticoagulation<br />
with warfarin<br />
10 ADEs in<br />
dispensing<br />
medications<br />
11 ADEs in drug<br />
dispensing <strong>and</strong>/or<br />
administration<br />
12 Hospital-acquired<br />
infections<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
Computerized<br />
physician order<br />
entry (CPOE) with<br />
clinical decision<br />
support system<br />
(CDSS)<br />
Clinical pharmacist<br />
consultation<br />
services<br />
Use <strong>of</strong> computer<br />
monitoring for<br />
potential ADEs<br />
Protocols for high<br />
risk drugs:<br />
nomograms for<br />
heparin<br />
Anticoagulation<br />
services <strong>and</strong> clinics<br />
for coumadin 8<br />
<strong>Patient</strong> selfmanagement<br />
using<br />
home monitoring<br />
devices<br />
Unit-dosing<br />
distribution system<br />
Use <strong>of</strong> automated<br />
medication<br />
dispensing devices<br />
Improved<br />
h<strong>and</strong>washing<br />
compliance (via<br />
education/behavior<br />
change; sink<br />
technology <strong>and</strong><br />
placement;<br />
Impact Study<br />
Strength<br />
633<br />
Effect Size Vigilance Cost Complexity<br />
High Medium 1 Modest Medium High 2<br />
High Medium Modest 3<br />
Medium Medium Robust 4<br />
High<br />
Low High Low<br />
Low Medium 5<br />
Low<br />
Medium Medium 6 Robust 7 Medium Low Low<br />
High Medium Unclear Low Medium Low<br />
High High Robust Medium Medium 9 High 10<br />
Medium<br />
11<br />
High Medium<br />
Medium Unclear Low Low Low<br />
12<br />
High Medium<br />
14<br />
Unclear Medium Medium<br />
Unclear 15<br />
13<br />
Low<br />
Low Low Low 16
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
13 Serious<br />
nosocomial<br />
infections (eg,<br />
vancomycinresistant<br />
enterococcus, C.<br />
difficile)<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
washing substance)<br />
Barrier precautions<br />
(via gowns &<br />
gloves; dedicated<br />
equipment;<br />
dedicated<br />
personnel)<br />
14 Hospital-acquired Limitations placed<br />
infections due to<br />
antibiotic-resistant<br />
organisms<br />
on antibiotic use<br />
15.1 Hospital-acquired<br />
urinary tract<br />
infection<br />
15.2 Hospital-acquired<br />
urinary tract<br />
infection<br />
16.1 Central venous<br />
catheter-related<br />
blood infections<br />
16.2 Central venous<br />
catheter-related<br />
blood infections<br />
16.3 Central venous<br />
catheter-related<br />
blood infections<br />
16.4 Central venous<br />
catheter-related<br />
blood infections<br />
16.4 Central venous<br />
catheter-related<br />
blood infections<br />
16.4 Central venous<br />
catheter-related<br />
blood infections<br />
Use <strong>of</strong> silver alloycoated<br />
catheters<br />
Use <strong>of</strong> suprapubic<br />
catheters<br />
Use <strong>of</strong> maximum<br />
sterile barriers<br />
during catheter<br />
insertion<br />
Antibioticimpregnated<br />
catheters<br />
Cleaning site<br />
(povidone-iodine<br />
to chlorhexidine)<br />
Changing catheters<br />
routinely<br />
Impact Study<br />
Strength<br />
High Medium<br />
634<br />
17<br />
Effect Size Vigilance Cost Complexity<br />
Robust Medium<br />
High 20 Medium Modest Medium<br />
High High Unclear 22<br />
High High Unclear<br />
23<br />
18<br />
21<br />
Medium Low 19<br />
Low Low<br />
Low Low Low<br />
Medium High High<br />
Medium High Robust Low Low Low 24<br />
Medium High Robust Low 25<br />
Low Low<br />
Medium High Unclear Low Low Low<br />
Medium High Negligible<br />
±<br />
NA High High<br />
Use <strong>of</strong> heparin Medium High Unclear Medium Low Low<br />
Tunneling shortterm<br />
central venous<br />
catheters<br />
± Actually, studies show a detrimental effect <strong>of</strong> practice.<br />
Medium High Unclear Low Low High
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
16.4 Central venous<br />
catheter-related<br />
blood infections<br />
17.1 Ventilatorassociated<br />
pneumonia<br />
17.1 Ventilatorassociated<br />
pneumonia<br />
17.2 Ventilatorassociated<br />
pneumonia<br />
17.3 Ventilatorassociated<br />
pneumonia<br />
17.4 Ventilatorassociated<br />
pneumonia<br />
18 Mortality<br />
associated with<br />
surgical<br />
procedures<br />
20.1 Surgical site<br />
infections<br />
20.2 Surgical site<br />
infections<br />
20.3 Surgical site<br />
infections<br />
20.4 Surgical site<br />
infections<br />
21 Morbidity due to<br />
central venous<br />
catheter insertion<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
Routine antibiotic<br />
prophylaxis<br />
Semi-recumbent<br />
positioning<br />
Continuous<br />
oscillation<br />
Continuous<br />
aspiration <strong>of</strong><br />
subglottic<br />
secretions (CASS)<br />
Selective<br />
decontamination <strong>of</strong><br />
digestive tract<br />
Impact Study<br />
Strength<br />
635<br />
Effect Size Vigilance Cost Complexity<br />
Medium Medium Negligible Medium Medium Low<br />
High Medium Robust 26<br />
Low Low Low<br />
High High Robust 27 Medium Medium Low<br />
High High Robust 28<br />
High High Robust 30 Medium<br />
Sucralfate High High Unclear High 32<br />
Localizing specific<br />
surgeries <strong>and</strong><br />
procedures to high<br />
volume centers<br />
Appropriate use <strong>of</strong><br />
antibiotic<br />
prophylaxis<br />
Maintenance <strong>of</strong><br />
perioperative<br />
normothermia<br />
Use <strong>of</strong><br />
supplemental<br />
perioperative<br />
oxygen<br />
Perioperative<br />
glucose control<br />
Use <strong>of</strong> real-time<br />
ultrasound<br />
guidance during<br />
central line<br />
insertion<br />
22 Surgical items left Counting sharps,<br />
instruments,<br />
High Medium<br />
Medium<br />
34<br />
33<br />
High Medium<br />
Low Low High 29<br />
31<br />
Low Low<br />
Low Low<br />
Varies Medium Varies High<br />
High Robust Medium<br />
36<br />
High Medium<br />
38<br />
35<br />
Robust Medium<br />
37<br />
Low Low<br />
Low Low<br />
Robust Low Low Low<br />
High Medium Robust Medium Low High 39<br />
High High Robust 40<br />
Low 41 Medium High<br />
Insuff. Low Not rated Not rated Low Low
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
inside patient sponges Info. 42<br />
23 Complications Use <strong>of</strong> preoperative<br />
due to anesthesia anesthesia<br />
equipment failures checklists<br />
24 <strong>Critical</strong> events in<br />
anesthesia<br />
25 Perioperative<br />
cardiac events in<br />
patients<br />
undergoing<br />
noncardiac<br />
surgery<br />
Intraoperative<br />
monitoring <strong>of</strong> vital<br />
signs <strong>and</strong><br />
oxygenation<br />
Use <strong>of</strong><br />
perioperative betablockers<br />
26.1 Falls Use <strong>of</strong><br />
identification<br />
bracelets<br />
26.2 Restraint-related<br />
injuries; Falls<br />
Interventions to<br />
reduce the use <strong>of</strong><br />
physical restraints<br />
safely<br />
Impact Study<br />
Strength<br />
Low 43<br />
Low 44 Medium<br />
45<br />
636<br />
Effect Size Vigilance Cost Complexity<br />
Low Not rated Not rated Low Low<br />
Unclear 46<br />
Low Low Low<br />
High High Robust Medium Low Low<br />
Medium Medium Negligible Low Low Low<br />
Medium Medium Unclear 47 Medium Medium Low<br />
26.3 Falls Use <strong>of</strong> bed alarms Medium Medium Unclear Low 48 Medium<br />
26.4 Falls <strong>and</strong> fallrelated<br />
injuries<br />
26.5 Falls <strong>and</strong> fall<br />
injuries<br />
Use <strong>of</strong> special<br />
flooring material in<br />
patient care areas<br />
Use <strong>of</strong> hip<br />
protectors<br />
27 Pressure ulcers Use <strong>of</strong> pressure<br />
relieving bedding<br />
materials<br />
28 Hospital-related<br />
delirium<br />
Multi-component<br />
delirium<br />
prevention<br />
program<br />
29 Hospital-acquired Geriatric<br />
complications (eg, consultation<br />
falls, delirium,<br />
functional decline,<br />
mortality)<br />
services<br />
49<br />
Low<br />
Medium Low Not rated Not rated High Low<br />
Medium High Robust Medium Low 50<br />
High High Robust 52<br />
Low 51<br />
Low High Low<br />
High Medium Robust Low Medium Low<br />
High High Varies 53<br />
Low Medium High
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
30 Hospital-acquired Geriatric<br />
complications evaluation <strong>and</strong><br />
(functional<br />
decline, mortality)<br />
management unit<br />
31 Venous<br />
thromboembolism<br />
(VTE)<br />
32 Contrast-induced<br />
renal failure<br />
32 Contrast-induced<br />
renal failure<br />
32 Contrast-induced<br />
renal failure<br />
33 Morbidity <strong>and</strong><br />
mortality in postsurgical<br />
<strong>and</strong><br />
critically ill<br />
patients<br />
34 Stress-related<br />
gastrointestinal<br />
bleeding<br />
35 Clinically<br />
significant<br />
misread<br />
radiographs <strong>and</strong><br />
CT scans by nonradiologists<br />
36 Pneumococcal<br />
pneumonia<br />
37.1 Inadequate pain<br />
relief in hospital<br />
patients with<br />
abdominal pain<br />
37.2 Inadequate pain<br />
relief<br />
Appropriate VTE<br />
prophylaxis<br />
Use <strong>of</strong> low osmolar<br />
contrast media<br />
Hydration<br />
protocols with<br />
theophylline<br />
Hydration<br />
protocols with<br />
acetylcysteine<br />
Various nutritional<br />
strategies<br />
Impact Study<br />
Strength<br />
High High Modest 54<br />
637<br />
Effect Size Vigilance Cost Complexity<br />
Low Medium High<br />
High High Robust Medium Low Low 55<br />
Medium High Robust Low High 56<br />
Low<br />
Medium High Negligible Low Low Low<br />
Medium Medium<br />
57<br />
Robust Low Low Low<br />
High High Robust 58 Medium Medium Low<br />
H2-antagonists Medium High Unclear Medium<br />
59<br />
Education<br />
interventions <strong>and</strong><br />
continuous quality<br />
improvement<br />
strategies<br />
Methods to<br />
increase<br />
pneumococcal<br />
vaccination rate<br />
Use <strong>of</strong> analgesics<br />
in patients with<br />
acute abdomen<br />
without<br />
compromising<br />
diagnostic accuracy<br />
Low Low<br />
Medium Medium Robust Low Low Low<br />
Medium High Unclear 60 Low 61<br />
High Medium<br />
Acute pain service High Medium Robust 64<br />
62<br />
Low Low<br />
Robust Medium Low Low 63<br />
Low 65 Medium Low
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
37.4 Inadequate<br />
postoperative pain<br />
management<br />
38 Morbidity <strong>and</strong><br />
mortality in ICU<br />
patients<br />
39 Morbidity <strong>and</strong><br />
mortality<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
Nonpharmacologic<br />
interventions (eg,<br />
relaxation,<br />
distraction)<br />
Change in ICU<br />
structure—active<br />
management by<br />
intensivist<br />
Changes in nursing<br />
staffing<br />
40 Any safety Promoting a<br />
problem amenable culture <strong>of</strong> safety<br />
to culture<br />
41.1 <strong>Medical</strong> device<br />
related adverse<br />
events<br />
Use <strong>of</strong> human<br />
factors principles<br />
in evaluation <strong>of</strong><br />
medical devices<br />
41.2 Adverse events Refining<br />
performance <strong>of</strong><br />
medical device<br />
alarms (eg,<br />
balancing<br />
sensitivity <strong>and</strong><br />
specificity <strong>of</strong><br />
alarms, ergonomic<br />
design)<br />
42.1 Adverse events<br />
related to<br />
discontinuities in<br />
care<br />
42.2 Adverse events<br />
during crosscoverage<br />
Information<br />
transfer between<br />
inpatient <strong>and</strong><br />
outpatient<br />
pharmacy<br />
St<strong>and</strong>ardized,<br />
structured signouts<br />
for physicians<br />
42.3 Adverse events Use <strong>of</strong> structured<br />
related to discharge<br />
information loss at summaries<br />
discharge<br />
42.4 Failures to<br />
communicate<br />
significant<br />
abnormal results<br />
Protocols for<br />
notification <strong>of</strong> test<br />
results to patients<br />
Impact Study<br />
Strength<br />
638<br />
Effect Size Vigilance Cost Complexity<br />
High High Unclear Low Low Low<br />
High Medium Robust 66<br />
High Medium<br />
Insuff.<br />
Info.<br />
Insuff.<br />
Info.<br />
High 71<br />
67<br />
** 69<br />
** 70<br />
** 72<br />
Low Medium High<br />
Varies Low High Low 68<br />
High Medium Robust Low Medium<br />
Medium Low Not rated Not rated Low 74<br />
Insuff.<br />
Info<br />
Varies High<br />
Varies High<br />
Varies High<br />
73<br />
Low<br />
Low<br />
Low 75 Not rated Not rated Low Low<br />
Medium Medium Modest Low Low Low
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
(eg, pap smears)<br />
43.1 Adverse events<br />
due to patient<br />
misidentification<br />
43.2 Performance <strong>of</strong><br />
invasive<br />
diagnostic or<br />
therapeutic<br />
procedure on<br />
wrong body part<br />
44 Adverse events<br />
related to team<br />
performance<br />
issues<br />
45 Adverse events<br />
due to provider<br />
inexperience or<br />
unfamiliarity with<br />
certain procedures<br />
<strong>and</strong> situations<br />
46 Adverse events<br />
related to fatigue<br />
in health care<br />
workers<br />
46 Adverse events<br />
related to fatigue<br />
in health care<br />
workers<br />
46 Adverse events<br />
related to fatigue<br />
in health care<br />
workers<br />
47 Adverse events<br />
due to<br />
transportation <strong>of</strong><br />
critically ill<br />
patients between<br />
health care<br />
facilities<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
Use <strong>of</strong> bar coding High 76<br />
“Sign your site”<br />
protocols<br />
Application <strong>of</strong><br />
aviation style crew<br />
resource<br />
management (eg,<br />
Anesthesia Crisis<br />
Management;<br />
MedTeams)<br />
Simulator-based<br />
training<br />
Limiting individual<br />
provider’s hours <strong>of</strong><br />
service<br />
Fixed shifts or<br />
forward shift<br />
rotations<br />
Impact Study<br />
Strength<br />
639<br />
Effect Size Vigilance Cost Complexity<br />
Low Not rated Not rated Varies 77<br />
High<br />
High Low Not rated Not rated Low High<br />
High 78<br />
Insuff.<br />
Info 79<br />
Insuff.<br />
Info.<br />
Insuff.<br />
Info.<br />
Napping strategies Insuff.<br />
Info.<br />
Specialized teams<br />
for interhospital<br />
transport<br />
Low Not rated Not rated Medium High<br />
Medium<br />
80<br />
Unclear<br />
81<br />
Low Medium Low<br />
Medium Unclear Low High High<br />
** 82<br />
** 84<br />
Medium Medium<br />
86<br />
Varies<br />
83<br />
High 85<br />
Varies<br />
Low<br />
Modest Low Medium Low
Ch. # <strong>Patient</strong> <strong>Safety</strong><br />
Target<br />
47 Adverse events<br />
due to<br />
transportation <strong>of</strong><br />
critically ill<br />
patients within a<br />
hospital<br />
48 Missed,<br />
incomplete or not<br />
fully<br />
comprehended<br />
informed consent<br />
48 Missed,<br />
incomplete or not<br />
fully<br />
comprehended<br />
informed consent<br />
48 Missed,<br />
incomplete or not<br />
fully<br />
comprehended<br />
informed consent<br />
49 Failure to honor<br />
patient<br />
preferences for<br />
end-<strong>of</strong>-life care<br />
49 Failure to honor<br />
patient<br />
preferences for<br />
end-<strong>of</strong>-life care<br />
<strong>Patient</strong> <strong>Safety</strong><br />
Practice<br />
Mechanical<br />
ventilation<br />
Asking that<br />
patients recall <strong>and</strong><br />
restate what they<br />
have been told<br />
during informed<br />
consent<br />
Use <strong>of</strong> video or<br />
audio stimuli<br />
Provision <strong>of</strong><br />
written informed<br />
consent<br />
information<br />
Computergenerated<br />
reminders to<br />
discuss advanced<br />
directives<br />
Use <strong>of</strong> physician<br />
order form for lifesustaining<br />
treatment (POLST)<br />
Impact Study<br />
Strength<br />
640<br />
Effect Size Vigilance Cost Complexity<br />
Medium Medium Negligible Low Low Low<br />
High Medium Robust Low Low Low 87<br />
High Medium Modest Low Low 88<br />
Low<br />
High Medium Unclear Low Low Low<br />
High Medium Robust Low Medium<br />
89<br />
Low<br />
High Low Not rated Not rated Low Low 90
Comments Section<br />
1 Medium strength <strong>of</strong> evidence for computerized physician order entry: although r<strong>and</strong>omized<br />
control trials have been conducted, findings from sophisticated “home grown” systems only 2-3<br />
sites may not be fully generalizable. In addition, the impact <strong>of</strong> the practice on adverse events has<br />
not been as well studied as for the non-clinical outcome, medication errors.<br />
2 Cost <strong>of</strong> CPOE is substantially higher than for most other practices in the high cost category.<br />
3 The impact <strong>of</strong> clinical pharmacists consultation services may be less than that <strong>of</strong> CPOE due to<br />
logistics <strong>of</strong> screening large volumes <strong>of</strong> orders to target those most prone to error or most<br />
consequential.<br />
4 Estimate <strong>of</strong> effect size based on single study with limited target (only antibiotic treatments).<br />
5 Cost influenced by whether existing computer systems are used in pharmacy services.<br />
6<br />
For nomogram protocols, study strength medium because the major concern (bleeding) is not<br />
addressed in most studies.<br />
7 Effect size greater than 15% for surrogate markers; not bleeding or clot rate.<br />
8<br />
Anticoagulation clinics: Both inpatient <strong>and</strong> outpatient venues studied, so some heterogeneity<br />
among results.<br />
9 Self-management <strong>of</strong> warfarin (coumadin): on average the cost per patient is low, but the<br />
aggregated cost is medium from the perspective <strong>of</strong> an insurer or integrated system.<br />
10 Higher complexity <strong>of</strong> implementation because self-management practice displaces locus <strong>of</strong><br />
control out <strong>of</strong> institution <strong>and</strong> may engender debate over insurance coverage. Other countries<br />
cover this practice, but it is currently not covered by Medicare in the United States.<br />
11<br />
Unit-dosing is a ubiquitous practice that has surprisingly little evidence <strong>of</strong> effectiveness;<br />
evidence is old <strong>and</strong> mixed.<br />
12<br />
Study strength is affected because outcomes measured are not the major outcomes <strong>of</strong> interest -<br />
ie, ADEs.<br />
13 The implementation ratings are related to patient safety only, but note that institutions may<br />
also implement this practice for cost-savings due to less drug loss <strong>and</strong> better inventory control.<br />
14<br />
Study design for h<strong>and</strong>washing compliance practices generally had short duration <strong>of</strong> follow-up;<br />
no r<strong>and</strong>omized control trials.<br />
641
15<br />
Unclear effect size due to mixed results <strong>and</strong> no clear pattern in a group <strong>of</strong> heterogeneous<br />
practices.<br />
16 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
17<br />
There are a number <strong>of</strong> studies <strong>of</strong> barrier precautions, but most are Level 3 study designs so the<br />
strength is not rated as “High.”<br />
18 Potential decrease in provider interaction with patients may cause psychological, as well as<br />
other, effects if care from clinicians is compromised.<br />
19 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
20<br />
Impact upgraded from “medium” to “high” rating because <strong>of</strong> public health impact <strong>of</strong> more<br />
antibiotic-resistant pathogens.<br />
21 Practice requires active, ongoing monitoring <strong>and</strong> input from infection control <strong>of</strong>ficers to make<br />
sure proper drugs are prescribed. Also, vigilance includes need for institution-wide monitoring <strong>of</strong><br />
pathogens.<br />
22 The effect size <strong>of</strong> using silver alloy catheters is unclear: a well-done meta-analysis is positive,<br />
showing decrease in bacteriuria, but more recent results <strong>of</strong> possibly better designed individual<br />
studies are mixed regarding benefit. Also, the actual strength <strong>of</strong> the link, however intuitive,<br />
between bacteriuria <strong>and</strong> clinically significant urinary tract infection is unclear.<br />
23 Effect size <strong>of</strong> using suprapubic catheters is unclear because <strong>of</strong> some heterogeneity in studies.<br />
Results are generally positive, but no meta-analysis yet conducted. In addition, the effect on<br />
outcome <strong>of</strong> clinically significant urinary tract infections is also unclear.<br />
24 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
25<br />
With antibiotic-impregnated catheters made with minocycline, there is the theoretical risk <strong>of</strong><br />
increased antibiotic resistance.<br />
26 Pneumonia outcome was significantly reduced, but mortality was not.<br />
27 Meta-analysis <strong>of</strong> 6 r<strong>and</strong>omized controlled trials showed significant <strong>and</strong> large relative risk<br />
reduction, but 2 other r<strong>and</strong>omized controlled trials showed no impact.<br />
28<br />
Benefit observed in prevention <strong>of</strong> ventilator-acquired pneumonia; no established benefit for<br />
mortality.<br />
642
29 High complexity for implementation since it requires retraining for a new practice.<br />
30 Most benefit in reducing pneumonia <strong>and</strong> mortality occurs when both IV <strong>and</strong> topical<br />
decontamination are used. Topical (by itself) only reduces ventilator-associated pneumonia.<br />
However, topical carries less potential for harm (ie, antibiotic resistance).<br />
31 Medium vigilance for harm because <strong>of</strong> public health concerns due to possible increase in<br />
antibiotic resistance. The Center for Disease Control <strong>and</strong> Prevention (CDC) <strong>and</strong> the American<br />
Thoracic Society (ATS) both recently reviewed this topic <strong>and</strong> did not recommend this practice.<br />
32<br />
If sucralfate were used because <strong>of</strong> its possible effect on reducing risk <strong>of</strong> ventilator-acquired<br />
pneumonia, it would displace a practice that has more established benefit for GI bleeding (H2<br />
blockers).<br />
33 The study strength for localizing care to high volume centers is evaluated across a range <strong>of</strong><br />
practices. There are large variations in evidentiary base across specific practices. Evidence is not<br />
structured to determine effect on patient safety. Although the literature includes possible<br />
benchmarks/thresholds for volume levels for specific procedures, the evidence is related more to<br />
quality enhancement than to improvements in patient safety.<br />
34 Relatively high current utilization <strong>of</strong> practice reduced impact by one level.<br />
35 Vigilance is required to monitor antibiotic overuse to prevent negative public health effects.<br />
36 Study strength is rated as medium because r<strong>and</strong>omized clinical trial data only applies to one<br />
disease process, although may be generalizable.<br />
37<br />
Medium vigilance for harm: although not studied, for certain cohorts the practice may be<br />
detrimental.<br />
38 Study strength is rated as medium because r<strong>and</strong>omized clinical trial data only applies to one<br />
disease process, although may be generalizable.<br />
39 Tight perioperative glucose control requires major shift in practice style, increased vigilance,<br />
more coordination between nurses <strong>and</strong> physicians, <strong>and</strong> perhaps new policies regarding nursing<br />
care for diabetics.<br />
40 Effect size high, but more impressive decrease in “failed insertion attempts” than in more<br />
clinically relevant complications. Also, there is some heterogeneity in study results, <strong>and</strong> there<br />
are two different technologies assessed (plain ultrasound vs. US with doppler), <strong>and</strong> the results<br />
vary.<br />
41 Theoretical risk that additional manipulation/h<strong>and</strong>ling could increase infection risk; also<br />
concern regarding impact on providers’ abilities to place catheters emergently when ultrasound<br />
guidance is not available.<br />
643
42 Insufficient information about retained sponges: the event is highly concerning <strong>and</strong> <strong>of</strong>ten<br />
morbid when occurs, but the only data on frequency are from case reports.<br />
43 Low potential impact because anesthesia complications are already so uncommon; also<br />
difficult to determine impact <strong>of</strong> current use <strong>of</strong> some version <strong>of</strong> this practice (eg, low opportunity<br />
possible due to current utilization).<br />
44 Low potential impact because anesthesia complications are already so uncommon; also<br />
difficult to determine impact <strong>of</strong> current use <strong>of</strong> some version <strong>of</strong> this practice (eg, low opportunity<br />
possible due to current utilization).<br />
45 Although there has been a very large r<strong>and</strong>omized trial <strong>of</strong> pulse oximetry, other studies covered<br />
additional aspects <strong>of</strong> intraoperative monitoring <strong>and</strong> were generally <strong>of</strong> lower study design quality.<br />
46 Pulse oximetry study showed no benefit, but major potential methodologic problems, such as<br />
secular trends. Complications that monitoring are designed to find are very unusual, so even a<br />
large trial may have been under-powered to detect important effects.<br />
47 Because the patient safety target is reduction <strong>of</strong> unnecessary restraints, there are multiple<br />
outcomes <strong>of</strong> interest. Although reducing unnecessary restraints does not seem to increase the risk<br />
<strong>of</strong> falls, it raises other concerns regarding disconnected IVs, elopement risk, etc., which have not<br />
been fully evaluated.<br />
48 Probably low, as categorized, but there is a theoretical risk that patients will not receive as<br />
much attention from nurses <strong>and</strong> other providers.<br />
49 Medium cost based on relatively widespread implementation <strong>of</strong> bed alarms required to impact<br />
all patients who may potentially benefit. May also impact nursing workload <strong>and</strong> staffing needs.<br />
50 Possibly higher cost if large numbers <strong>of</strong> patients would benefit from wearing hip protectors.<br />
There is also the question <strong>of</strong> whether these costs are borne by system/insurers or patients<br />
themselves.<br />
51 Implementation complexity in the hospital may be low, but implementation outside <strong>of</strong> the<br />
hospital might involve large educational campaign directed at patients who could benefit from<br />
practice.<br />
52 Studies compare a variety <strong>of</strong> special bedding materials to st<strong>and</strong>ard beds. Effect size for one<br />
special bed option versus another is not known. Unclear which particular surface works best.<br />
53 Effect size varies since heterogeneous outcomes, perhaps in part related to the variety <strong>of</strong><br />
interventions, some <strong>of</strong> which involved both inpatient <strong>and</strong> outpatient components.<br />
54 Effect size varies due to heterogeneous results, which depend in part on the outcomes <strong>of</strong><br />
interest (ie, functional outcomes vs. mortality).<br />
644
55 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
56 Total cost, <strong>of</strong> course, depends on the extent <strong>of</strong> utilization (eg, all patients versus only targeted<br />
patients). Cost-effectiveness analyses demonstrate the importance <strong>of</strong> targeting appropriate<br />
patients.<br />
57 Outcome is level 2, only one study for N-acetylcystine.<br />
58 Varies according to specific nutritional support practice. Robust findings for early enteral<br />
nutrition in critically ill <strong>and</strong> post-surgical patients.<br />
59 Vigilance for harm is medium because <strong>of</strong> potential risk <strong>of</strong> increasing ventilator-associated<br />
pneumonia, <strong>and</strong> also because <strong>of</strong> possible overuse since high-risk groups are now better defined.<br />
60 Depends on specific intervention; st<strong>and</strong>ing orders have the highest effectiveness.<br />
61<br />
Harm concern low, except one recent study (see Chapter 36) showed trend toward harm in<br />
HIV-positive patients.<br />
62 Although some studies were r<strong>and</strong>omized control trials, they did not look at all clinically<br />
relevant outcomes to ensure that practice was safe. Under-powered to assess whether diagnostic<br />
capability not impaired.<br />
63 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
64 Only studied for post-operative pain; may not apply more generally.<br />
65<br />
Some speculation that care may be fragmented when applied broadly, beyond post-operative<br />
patients.<br />
66 Some <strong>of</strong> the positive results may be attributable to factors other than the intervention. Concern<br />
about underlying population changing (eg, secular trends).<br />
67 Study strength is medium despite a number <strong>of</strong> studies, because <strong>of</strong> variation in practices (eg,<br />
various measures <strong>of</strong> nurse staffing, models <strong>of</strong> care). Chapter was designed to generalize across<br />
practices regarding nursing structure versus outcomes; evidence is not structured to tell effect on<br />
patient safety, <strong>and</strong> there are no benchmarks/thresholds for nurse staffing levels.<br />
68 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
645
69 Most evidence available outside <strong>of</strong> medicine; study strength not rated. These practices, drawn<br />
largely from non-health care industries, were not fully rated because <strong>of</strong> their unique nature <strong>and</strong><br />
their relatively small evidentiary base in the health care literature.<br />
70 Most evidence available outside <strong>of</strong> medicine; study strength not rated. These practices, drawn<br />
largely from non-health care industries, were not fully rated because <strong>of</strong> their unique nature <strong>and</strong><br />
their relatively small evidentiary base in the health care literature.<br />
71 Although alarms are ubiquitous in the hospital, it is unclear how many adverse events might<br />
be improved by improvements in alarm systems.<br />
72 Most evidence available outside <strong>of</strong> medicine; study strength not rated. These practices, drawn<br />
largely from non-health care industries, were not fully rated because <strong>of</strong> their unique nature <strong>and</strong><br />
their relatively small evidentiary base in the health care literature.<br />
73 Costs are shared among a variety <strong>of</strong> payors including outpatient pharmacy.<br />
74 Cost would vary based on interventions considered—some low-tech, paper-based, or pocket<br />
computers; higher cost for full-scale computerized systems.<br />
75<br />
Although one r<strong>and</strong>omized trial performed, the outcomes reported were only indirectly related<br />
to patient safety outcomes.<br />
76 Somewhat unclear, but errors due to misidentification can be grave.<br />
77 Cost varies based on specific system <strong>and</strong> level <strong>of</strong> implementation.<br />
78<br />
Impact is a function <strong>of</strong> how widely the practice can be used (ICU vs. ward teams vs. operating<br />
room).<br />
79<br />
Insufficient information outside <strong>of</strong> anesthesia about volume <strong>of</strong> human factors errors amenable<br />
to training approaches.<br />
80<br />
Limited studies with small numbers <strong>of</strong> participant <strong>and</strong> with different simulators lead to<br />
concerns about generalizability.<br />
81<br />
Effect unclear since few studies with comparable simulators, <strong>and</strong> evaluated with mostly Level<br />
3 outcomes.<br />
82 Most evidence available outside <strong>of</strong> medicine; study strength not rated. These practices, drawn<br />
largely from non-health care industries, were not fully rated because <strong>of</strong> their unique nature <strong>and</strong><br />
their relatively small evidentiary base in the health care literature.<br />
83 Fixed shift may be more costly <strong>and</strong> difficult to implement than forward rotation.<br />
646
84 Most evidence available outside <strong>of</strong> medicine; study strength not rated. These practices, drawn<br />
largely from non-health care industries, were not fully rated because <strong>of</strong> their unique nature <strong>and</strong><br />
their relatively small evidentiary base in the health care literature.<br />
85<br />
Restructuring patient care to allow for napping while minimizing discontinuities could be<br />
expensive.<br />
86 Study strength is borderline-medium with three Level 3 studies.<br />
87 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
88 Cost for video disks – assumes that <strong>of</strong>f-the-shelf products exist for common procedures; would<br />
be higher if an institution has to build its own systems.<br />
89<br />
Cost would be lower for health care organizations that already rely on computers for care<br />
management.<br />
90 Rated as low, but this practices requires behavior change on the part <strong>of</strong> the provider.<br />
Therefore, it may be more difficult to implement because its success largely rests on education<br />
(see Chapter 54) <strong>and</strong> acceptance.<br />
647
648
Appendix: List <strong>of</strong> Contributors<br />
Editorial Board<br />
Kaveh G. Shojania, MD<br />
Assistant Pr<strong>of</strong>essor<br />
Department <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco, CA<br />
Bradford W. Duncan, MD<br />
AHRQ <strong>Patient</strong> <strong>Safety</strong> Fellow<br />
Center for Primary Care & Outcomes<br />
Research<br />
Stanford University<br />
Stanford, CA<br />
649<br />
Kathryn M. McDonald, MM<br />
Executive Director<br />
Center for Primary Care & Outcomes<br />
Research<br />
Stanford University<br />
Stanford, CA<br />
Robert M. Wachter, MD<br />
Pr<strong>of</strong>essor <strong>and</strong> Associate Chairman<br />
Department <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco, CA
Contributors<br />
Joseph V. Agostini, MD<br />
Robert Wood Johnson Clinical Scholar<br />
Yale University School <strong>of</strong> Medicine<br />
New Haven, CT<br />
Chapters 26-30<br />
Andrew D. Auerbach, MD, MPH<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco, CA<br />
Chapters 18-20, 22-25<br />
Dorothy I. Baker, PhD, RNCS<br />
Research Scientist, Chronic Disease Epidemiology<br />
Department <strong>of</strong> Epidemiology <strong>and</strong> <strong>Public</strong> Health<br />
Associate Director, Claude Pepper Older Americans<br />
Independence Center<br />
Yale University School <strong>of</strong> Medicine<br />
New Haven, CT<br />
Chapters 26-30<br />
David W. Bates, MD, MSc<br />
Associate Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Harvard <strong>Medical</strong> School<br />
Chief, Division <strong>of</strong> General Internal Medicine<br />
Brigham <strong>and</strong> Women's Hospital<br />
Boston, MA<br />
Chapters 6-9, 41, 42, 45, <strong>and</strong> 46<br />
Sidney T. Bogardus, Jr., MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Yale University School <strong>of</strong> Medicine<br />
<strong>Medical</strong> Director, Adler Geriatric Assessment<br />
Center<br />
Yale-New Haven Hospital<br />
New Haven, CT<br />
Chapters 26-30<br />
650<br />
Erica Brownfield, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapter 37<br />
Harold R. Collard, MD<br />
Fellow in Pulmonary & <strong>Critical</strong> Care Medicine<br />
University <strong>of</strong> Colorado Health Sciences Center<br />
Denver, CO<br />
Chapter 17<br />
Lorenzo Di Francesco, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Department <strong>of</strong> Medicine<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapter 32<br />
Daniel D. Dressler, MD<br />
Instructor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Hospital Medicine Fellow<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapter 34<br />
Bradford W. Duncan, MD<br />
AHRQ <strong>Patient</strong> <strong>Safety</strong> Fellow<br />
Center for Health Policy<br />
Center for Primary Care & Outcomes Research<br />
Stanford University<br />
Stanford, CA<br />
Chapters 45 <strong>and</strong> 46
Sharon K. Inouye, MD, MPH<br />
Associate Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Yale University School <strong>of</strong> Medicine<br />
Investigative Medicine Program<br />
New Haven, CT<br />
Chapters 28 <strong>and</strong> 29<br />
Salim D. Islam, MD<br />
Assistant Clinical Pr<strong>of</strong>essor<br />
Department <strong>of</strong> Anesthesia <strong>and</strong> Perioperative Care<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco VA <strong>Medical</strong> Center<br />
San Francisco, CA 94121<br />
Chapters 23 <strong>and</strong> 24<br />
Ashish K. Jha, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
Division <strong>of</strong> General Internal Medicine<br />
San Francisco VA <strong>Medical</strong> Center<br />
San Francisco, CA<br />
Chapters 45 <strong>and</strong> 46<br />
Rainu Kaushal, MD, MPH<br />
Instructor in Medicine<br />
Harvard <strong>Medical</strong> School<br />
Division <strong>of</strong> General Internal Medicine<br />
Brigham <strong>and</strong> Women’s Hospital<br />
Boston, MA<br />
Chapter 6 <strong>and</strong> 7<br />
651<br />
Jennifer Kleinbart, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Atlanta, GA<br />
Chapter 31<br />
Sunil Kripalani, MD<br />
Instructor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Division <strong>of</strong> General Medicine<br />
Grady Memorial Hospital<br />
Atlanta, Georgia<br />
Chapter 35<br />
Ebbing Lautenbach, MD, MPH, MSCE<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine <strong>and</strong><br />
Epidemiology<br />
University <strong>of</strong> Pennsylvania School <strong>of</strong> Medicine<br />
Hospital Epidemiologist, Presbyterian <strong>Medical</strong><br />
Center<br />
Philadelphia, PA<br />
Chapters 12-14<br />
Susana B. Martins, MD, MSc<br />
Fellow, Institute for Health Policy Studies<br />
University <strong>of</strong> California, San Francisco<br />
Foster City, CA<br />
Chapter 47
Harvey J. Murff, MD<br />
Fellow, Division <strong>of</strong> General Internal Medicine<br />
Brigham <strong>and</strong> Women's Hospital<br />
Boston, MA<br />
Chapters 23, 41 <strong>and</strong> 42<br />
Michael D. Murray, PharmD, MPH<br />
Pr<strong>of</strong>essor <strong>of</strong> Pharmacy<br />
Purdue University School <strong>of</strong> Pharmacy<br />
Scientist, Regenstrief Institute for Healthcare<br />
Regenstrief Health Center, Sixth Floor<br />
Indianapolis, Indiana<br />
Chapters 10 <strong>and</strong> 11<br />
David B. Nash, MD, MBA<br />
Associate Dean <strong>and</strong> Director<br />
Office <strong>of</strong> Health Policy <strong>and</strong> Clinical Outcomes<br />
Thomas Jefferson University<br />
Philadelphia, PA<br />
Chapters 40, 44, 48 <strong>and</strong> 50<br />
Steven Z. Pantilat, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
Project on Death in America Faculty Scholar<br />
San Francisco, CA<br />
Chapter 49<br />
652<br />
Laura T. Pizzi, PharmD<br />
Project Director<br />
Office <strong>of</strong> Health Policy <strong>and</strong> Clinical Outcomes<br />
Thomas Jefferson University<br />
Philadelphia, PA<br />
Chapters 40, 44, 48 <strong>and</strong> 50<br />
Jeffrey M. Rothschild, MD, MPH<br />
Instructor in Medicine<br />
Harvard <strong>Medical</strong> School<br />
Division <strong>of</strong> General Medicine<br />
Brigham <strong>and</strong> Women's Hospital<br />
Boston, MA<br />
Chapters 21 <strong>and</strong> 38<br />
Kimberly Rask, MD, PhD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine <strong>and</strong> Health Policy<br />
Emory University Schools <strong>of</strong> Medicine <strong>and</strong> <strong>Public</strong><br />
Health<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapters 31,34, <strong>and</strong> 35<br />
Sanjay Saint, MD, MPH<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Director, <strong>Patient</strong> <strong>Safety</strong> Enhancement Program<br />
University <strong>of</strong> Michigan School <strong>of</strong> Medicine<br />
Research Scientist<br />
Ann Arbor VA Health Services Research &<br />
Development Field Program<br />
Ann Arbor, MI<br />
Chapters 15-17
Jean Ann Seago, PhD, RN<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Nursing<br />
University <strong>of</strong> California, San Francisco<br />
Research Fellow<br />
Center for California Health Workforce Studies<br />
San Francisco, CA<br />
Chapter 39<br />
Kaveh G. Shojania, MD<br />
Assistant Pr<strong>of</strong>essor<br />
Department <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco, CA<br />
Chapters 4, 5, 9, 10, 43, <strong>and</strong> 47<br />
Robert Trowbridge, MD<br />
Clinical Instructor<br />
Department <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
San Francisco, CA<br />
Chapters 51-55<br />
Robert M. Wachter, MD<br />
Pr<strong>of</strong>essor <strong>and</strong> Associate Chairman<br />
Department <strong>of</strong> Medicine<br />
University <strong>of</strong> California, San Francisco<br />
Chief <strong>of</strong> the <strong>Medical</strong> Service<br />
M<strong>of</strong>fitt-Long Hospitals<br />
San Francisco, CA<br />
Chapter 55<br />
653<br />
Heidi Wald, MD<br />
Assistant Pr<strong>of</strong>essor in Medicine<br />
University <strong>of</strong> Pennsylvania<br />
Director, Penn Hospital Care Group<br />
Hospital <strong>of</strong> the University <strong>of</strong> Pennsylvania<br />
Philadelphia, PA<br />
Chapters 4,5 <strong>and</strong> 43<br />
Scott Weingarten, MD, MPH<br />
Associate Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
University <strong>of</strong> California, Los Angeles<br />
Director, Health Services Research<br />
Cedars-Sinai Health System<br />
Beverly Hills, CA<br />
Chapters 51-54<br />
Mark V. Williams, MD<br />
Associate Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Director, Hospital Medicine Unit<br />
Emory University School <strong>of</strong> Medicine<br />
<strong>Medical</strong> Director, Center for Clinical<br />
Effectiveness<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapters 31-35<br />
Neil Winawer, MD<br />
Assistant Pr<strong>of</strong>essor <strong>of</strong> Medicine<br />
Emory University School <strong>of</strong> Medicine<br />
Grady Memorial Hospital<br />
Atlanta, GA<br />
Chapter 33
654
Index<br />
abdominal pain, acute<br />
analgesics for, 396<br />
analgesics studies, 397, 399<br />
academic detailing for modifying physician behavior, 595,<br />
597<br />
accreditation methods to promote patient safety practices,<br />
607<br />
accreditation organizations, 602<br />
active error, 51<br />
acute pain services<br />
described, 400<br />
studies, 401, 403<br />
acute spinal cord injury, VTE risk, 338<br />
ADE detection<br />
alert monitors, 79<br />
computerized alert monitors, 79, 80<br />
computerized systems, studies, 83<br />
pharmacists, physicians, <strong>and</strong>, 79<br />
physician response to alert monitors, 81<br />
research needs, 82<br />
ADEs<br />
anticoagulants <strong>and</strong>, 87<br />
CDSS <strong>and</strong>, 590<br />
clinical pharmacists <strong>and</strong>, 71, 74, 75<br />
computerized detection <strong>and</strong> alert systems, 79<br />
described, 79<br />
incident reporting, 42<br />
information transfer between inpatient/outpatient<br />
pharmacies, 471<br />
prevalence, 71<br />
advance directives for end-<strong>of</strong>-life care, 556, 560<br />
Agency for Health Care Policy <strong>and</strong> Research (AHCPR),<br />
396, 406<br />
Agency for Healthcare Research <strong>and</strong> Quality (AHRQ)<br />
acknowledgments, 21<br />
patient advocacy, 569<br />
patient safety, 601<br />
patient safety improvement, 13<br />
American <strong>Medical</strong> Association (AMA) Code <strong>of</strong> <strong>Medical</strong><br />
Ethics, 546<br />
American Society for Health System Pharmacists (ASHP),<br />
antibiotic administration, 221<br />
American Society <strong>of</strong> Anesthesiologists (ASA), 259, 265,<br />
407, 604<br />
amphotericin, topical, 193<br />
analgesics<br />
acute abdominal pain management, 396<br />
acute abdominal pain management studies, 397, 399<br />
antiemetics for PCA, 404<br />
research needs, 398<br />
surgeon decision making , impact on, 398<br />
anesthesia<br />
body temperature <strong>and</strong>, 231<br />
intraoperative monitoring <strong>and</strong>, 265<br />
pre-anesthesia checklists, 259<br />
simulator-based training, 512<br />
anesthesia crisis resource management (ACRM), 504<br />
anesthesia equipment checklists, 466<br />
antibiotic resistance <strong>and</strong> pneumococcal vaccines, 390<br />
antibiotics<br />
655<br />
benefits for controlling nosocomial infections, 143<br />
practices to change use, 141<br />
prescription errors, 142<br />
prophylaxis for SSI, 221, 224, 226<br />
urinary catheters <strong>and</strong>, 149<br />
anticoagulation<br />
ADEs <strong>and</strong>, 88<br />
inpatient services, 94<br />
outpatient clinics, 88<br />
patient self-management, 89, 96<br />
research needs, 91<br />
antiemetics for pain management<br />
described, 404<br />
studies, 405<br />
ASEPSIS, 232, 235<br />
aspirin to prevent venous thromboembolism, 333<br />
atelectasis, 185, 234<br />
audit <strong>and</strong> feedback for modifying physician behavior, 595,<br />
596<br />
Australian Incident Monitoring Study (AIMS), 43, 45, 266<br />
bacteremia <strong>and</strong> urinary catheters, 149<br />
bacteriuria <strong>and</strong> urinary catheters, 149<br />
bar coding<br />
bedside scanning, 491<br />
described, 487<br />
medical recordkeeping, 489<br />
medication dispensing <strong>and</strong> administration, 488<br />
patient identification, 487<br />
patient identification studies, 490, 491<br />
research needs, 492<br />
specimen h<strong>and</strong>ling, 488<br />
barrier precautions, central venous catheter insertion<br />
contamination reduction, 165<br />
studies, 165<br />
barrier precautions, VRE <strong>and</strong> C. difficile<br />
compliance vs. specific intervention, 127<br />
described, 127<br />
further research, 130<br />
list <strong>of</strong>, 127<br />
studies, 129, 132, 135, 136<br />
Baxter ATC-212 dispensing system, 111<br />
bed alarms for fall prevention in older patients<br />
described, 291<br />
studies, 292, 293<br />
bed sores, 156<br />
beds or mattresses for pressure ulcer prevention<br />
described, 301<br />
studies, 302, 304<br />
bedsores, 301<br />
behavior-based safety interventions, 450<br />
Best Medicine, The, 570<br />
beta-blockers, perioperative<br />
described, 271<br />
research needs, 274<br />
studies, 272, 276<br />
bloodstream infections, catheter-related (CR-BSI)<br />
defined, 163<br />
risks, 164<br />
bronchopulmonary secretions, 185<br />
business community <strong>and</strong> patient safety, 603
calorimetry, 359<br />
C<strong>and</strong>ida species in catheter-related infections, 163<br />
capnography, 267<br />
cardiac disease, heparin <strong>and</strong> warfarin monitoring, 88<br />
cardiac events<br />
hypothermia, 231<br />
reducing with beta blockers, 271<br />
care delivery <strong>and</strong> interventions, nurse staffing <strong>and</strong>, 423,<br />
426, 430<br />
catheter colonization, 163<br />
catheter related infections<br />
central venous, 167<br />
catheter-related bloodstream infection (CR-BSI)<br />
antibacterial coatings <strong>and</strong> antiseptic agents, 169<br />
CHG solutions <strong>and</strong>, 175<br />
defined, 163, 164<br />
intravenous antimicrobial use <strong>of</strong>, 179<br />
catheter-related infections<br />
bloodstream, 163<br />
central venous, 164<br />
urinary tract, 149<br />
Centers for Disease Control <strong>and</strong> Prevention (CDC), 42,<br />
164, 194, 602<br />
antibiotic administration, 221<br />
pneumococcal vaccines, 385<br />
central venous catheters<br />
antibacterial coatings or antiseptic agents, 169, 174<br />
barrier precautions during insertion, 165<br />
CHG <strong>and</strong> PI studies, 177, 178<br />
chlorhexidine gluconate at the insertion site, 175<br />
heparin <strong>and</strong>, 178<br />
infection studies, 167<br />
placement <strong>of</strong>, 245<br />
prolonged use, infection risks <strong>and</strong> risk reduction, 165<br />
research needs, 183<br />
ultrasound guidance for insertion, 246<br />
use <strong>of</strong>, 163<br />
children, transporting, 535, 536, 537<br />
chlorhexidine gluconate (CHG) solution<br />
central venous catheter insertion, 175<br />
compared to PI, studies, 177, 178<br />
studies, 175<br />
chlorhexidine/silver sulfadiazine-impregnated catheters<br />
described, 169<br />
studies, 169, 172, 174<br />
circadian rhythm, 519, 524<br />
clinical decision support systems (CDSS)<br />
CPOE <strong>and</strong>, 589<br />
described, 589<br />
research needs, 593<br />
safety practice promotion <strong>and</strong> implementation, 590<br />
studies, 590, 591<br />
clinical pharmacists<br />
ADEs <strong>and</strong>, 71<br />
ambulatory settings, study results, 73<br />
consultation services, 71<br />
impact on ADEs, studies, 75<br />
outpatient settings, studies, 72<br />
roles <strong>and</strong> activities, 71<br />
closed ICU model, 413<br />
Clostridium difficile (C. difficile)<br />
antibiotic studies, 142, 143<br />
656<br />
antibiotics <strong>and</strong>, 141<br />
barrier precaution studies, 129<br />
barrier precautions <strong>and</strong>, 127<br />
relationship to VRE, 128<br />
research needs, 131, 144<br />
SSI <strong>and</strong>, 222<br />
transferability, 127<br />
Cockpit Management Attitudes Questionnaire (CMAQ),<br />
502<br />
computerized discharge summaries, 479<br />
computerized physician order entry (CPOE) with CDSS<br />
CDSS described, 589<br />
computerized sign-out program, 476, 479<br />
continuous intravenous insulin infusion (CII) during<br />
surgery, 238<br />
continuous oscillation<br />
described, 185, 187<br />
studies, 187, 189<br />
cranial computed tomography (CT), 376<br />
Crew Resource Management (CRM)<br />
aviation application vs. health care, 504<br />
aviation environment, 501<br />
described, 501<br />
effectiveness measurement tools, 502<br />
health care environment, 504<br />
operating room environment, 505<br />
peer monitoring, 505<br />
research needs, 507<br />
studies, 503, 506<br />
critical incident technique, 41<br />
critical pathways<br />
described, 581<br />
medical information dissemination, 581<br />
research needs, 584<br />
studies, 582, 585<br />
cross-coverage, 476<br />
Crossing the Quality Chasm: A New Health System for the<br />
21 st Century, 13<br />
decubitus ulcers, 156, 301<br />
deep vein thrombosis (DVT), 333, 341<br />
delirium in older hospitalized patients<br />
described, 307<br />
geriatric consultation services <strong>and</strong>, 313<br />
interventions, 307<br />
research needs, 309<br />
studies, 308, 310<br />
diabetes<br />
glucose control during surgery, 237<br />
RCIN <strong>and</strong>, 349<br />
diarrhea, C. difficile-associated, 128<br />
direct mail patient notification, 483<br />
discharge summaries<br />
computerized, 479<br />
research needs, 481<br />
structured vs. unstructured format, 479<br />
studies, 480, 481<br />
disseminating medical information, 581<br />
distal DVT, 333<br />
domains, defined, 29<br />
duplex ultrasonography, 334<br />
durable power <strong>of</strong> attorney for health care (DPOA-HC), 556<br />
educational techniques to modify physician behavior
described, 595<br />
research needs, 598<br />
studies, 596, 599<br />
elastic stockings (ES) to prevent venous thromboembolism,<br />
333<br />
elderly<br />
beta-blockers <strong>and</strong>, 273<br />
delirium, preventing, 307<br />
falling, preventing, 281<br />
falls, preventing, 283<br />
geriatric consultation services, 313<br />
pneumococcal infections, 386<br />
pneumococcal vaccines <strong>and</strong>, 387<br />
pressure ulcers, preventing, 301<br />
preventable deaths <strong>and</strong> functional decline, 324<br />
embolism to the pulmonary vasculature, 333<br />
end-<strong>of</strong>-life care<br />
administrative initiatives, 560<br />
advance directives, 560<br />
described, 556<br />
documenting patient preferences, 557<br />
existing programs/contributions, 562<br />
forms to promote implementation <strong>of</strong> patient preferences,<br />
559<br />
implementation <strong>of</strong> patient preferences, 557<br />
improving physician/patient communication, 560<br />
physician/patient communication, 558<br />
preferences in outpatient setttings, 560<br />
public education, 562<br />
research needs, 563<br />
enteral nutrition (EN)<br />
immune-enhanced, 363<br />
paths for administering, 359<br />
studies, 361<br />
environmental modifications to prevent delirium, 307<br />
equipment, anesthesia, checklists for, 259<br />
error mode prediction, 26<br />
examination equipment, dedicated or disposable, as a<br />
barrier precaution, 127, 136<br />
extended-spectrum beta-lactamase (ESBL), 128, 143<br />
falls in hospitalized <strong>and</strong> institutionalized older patients<br />
bed alarms <strong>and</strong>, 291<br />
geriatric consultation services <strong>and</strong>, 313<br />
hip protectors <strong>and</strong>, 295<br />
identification bracelets <strong>and</strong>, 284<br />
interventions, list <strong>of</strong>, 283<br />
physical restraints <strong>and</strong>, 287<br />
problem description, 281<br />
research needs, 283, 284, 289, 292, 294, 297<br />
risk assessment, 282<br />
risk factors, 281<br />
special flooring <strong>and</strong>, 293<br />
fatigue<br />
lighting at work, 524<br />
medical errors <strong>and</strong>, 519<br />
medical therapies, 525<br />
napping, 524<br />
research needs, 526<br />
shift limits <strong>and</strong> error reduction, 521<br />
shift rotation <strong>and</strong>, 520<br />
shift rotation, direction <strong>and</strong> speed, 523<br />
sleep hygiene education, 523<br />
657<br />
studies, 522<br />
fibrinogen leg scanning, 334<br />
Flight Management Attitudes Questionnaire (FMAQ), 502<br />
flooring, special, for fall prevention in older patients<br />
described, 293<br />
studies, 294, 295<br />
Foley catheters, 149<br />
Food <strong>and</strong> Drug Administration (FDA), 42, 51, 259, 262,<br />
459, 489<br />
gastroesophogeal reflux, 185<br />
gastrointestinal bleeding<br />
described, 368<br />
nosocomial pneumonia <strong>and</strong>, studies, 372<br />
research needs, 371<br />
studies, 369<br />
sucralfate <strong>and</strong>, 199<br />
gentamicin, topical, 193<br />
geriatric consultation services<br />
described, 313<br />
research needs, 316<br />
studies, 314, 315, 318<br />
geriatric evaluation <strong>and</strong> management (GEM) units<br />
described, 313, 323<br />
research needs, 326<br />
studies, 324, 327<br />
gloves/gowns as barrier precaution, 127<br />
glucose control, perioperative<br />
described, 237<br />
research needs, 240<br />
studies, 238, 241<br />
surgical infections <strong>and</strong>, 238<br />
government agencies <strong>and</strong> patient safety practices, 602<br />
gram negative rods in catheter-related infections, 163<br />
H2-receptor antagonists, 198, 368, 369, 372<br />
h<strong>and</strong> disinfection, 119<br />
h<strong>and</strong>washing<br />
compliance difficulties, 119<br />
relationship to infection, 119<br />
research needs, 122<br />
risk factors, 120<br />
skin irritation to health care workers, 121<br />
strategies to improve compliance, 121<br />
studies, 120, 123<br />
Health Care Financing Administration (HCFA)<br />
medical errors, 601<br />
physical restraints <strong>and</strong>, 288<br />
pneumococcal vaccines, 287, 385<br />
health literacy problem, 547<br />
heart failure<br />
ADEs <strong>and</strong>, 72<br />
clinical pharmacists <strong>and</strong>, 73<br />
heparin<br />
about, 87<br />
bleeding during therapy with, 88<br />
central venous catheters <strong>and</strong>, 178<br />
dosing <strong>and</strong> monitor errors, reducing, 87<br />
LDUH/LMWH to prevent venous thromboembolism,<br />
333<br />
monitoring, 88<br />
heparin nomograms<br />
described, 87<br />
PTTs <strong>and</strong>, 89
studies, 88, 92<br />
high reliability theory, 26, 447<br />
high-volume centers, localizing care to<br />
Leapfrog Group’s efforts, 208<br />
relationship to patient outcomes, 205<br />
research needs, 208<br />
studies, 206, 210<br />
hip fractures<br />
falls in older patients, 281<br />
VTE risk in surgery, 336<br />
hip protectors for fall prevention in older patients, 296, 298<br />
hip replacement surgery, VTE risk, 336<br />
histamine-2 receptor blocking agents (H2-receptor<br />
antagonists)<br />
described, 368<br />
studies, 369, 372<br />
VAP prevention, 198<br />
hospitalists <strong>and</strong> end-<strong>of</strong>-life care, 561<br />
human error identification, 26<br />
human factors engineering (HFE)<br />
anaesthesia equipment checklists, 466<br />
described, 459<br />
medical device alarms, 462<br />
medical devices <strong>and</strong> errors, 459<br />
research needs, 467<br />
hydration protocols to prevent RCIN, 349<br />
hygenic h<strong>and</strong> rub, 119<br />
hypercholesterolemia, clinical pharmacists <strong>and</strong>, 73<br />
hypertension, clinical pharmacists <strong>and</strong>, 73<br />
hypoxemia <strong>and</strong> anesthesia, 267<br />
identification bracelets for fall prevention in older patients<br />
study, 285<br />
using, 284<br />
incident reporting<br />
AIMS data vs. JCAHO data, 45<br />
barriers to linking data with outcome, 46<br />
critical incident monitoring, 41<br />
event appropriateness, 43<br />
intensive care units (ICUs), 42<br />
JCAHO surveillance m<strong>and</strong>ates, 43<br />
medication errors <strong>and</strong>, 41<br />
physician reporting, 45<br />
research needs, 46<br />
sentinel event alerts, 47<br />
industries outside health care<br />
aviation, 259, 501<br />
behavior-based safety programs, 450<br />
corporate cultures, 448<br />
critical pathways, 581<br />
disasters contributed to by fatigue, 520<br />
incident reporting systems, 41, 42<br />
safety cultures, 449<br />
simulator training, 511<br />
infection control <strong>of</strong>ficers, 602<br />
information transfer, improving<br />
abnormal results notification, 482<br />
discharge summaries, 479<br />
inpatient/outpatient pharmacies, 471<br />
research needs, 474, 486<br />
sign-out systems, 476<br />
informed consent<br />
658<br />
asking patient to recall discussion information, studies,<br />
550<br />
form readability studies, 549<br />
forms inadequacies, 548<br />
legal <strong>and</strong> ethical obligations, 546<br />
physician/patient discussion, studies, 549<br />
procedural inefficiences in obtaining, 547<br />
process, 546<br />
research needs, 552<br />
visual or auditory aids, studies, 550<br />
written information to patients, studies, 551<br />
Institute <strong>of</strong> Medicine (IOM) reports, 13, 205, 207, 210<br />
intensive care units (ICUs)<br />
incident reporting <strong>and</strong>, 42<br />
intensivist model studies, 418<br />
Leapgrog Group recommendations, 413<br />
nutritional support for patients, 359<br />
research needs, 416<br />
structure model studies, 414, 415<br />
structure models, 413<br />
intensivist co-management ICU model, 413<br />
intensivists in ICUs, 413<br />
interhospital transportation <strong>of</strong> critically ill patients, 536,<br />
537<br />
interhospital transportation <strong>of</strong> critically ill patients, 534<br />
intermittent pneumatic compression (IPC) to prevent<br />
venous thromboembolism, 333<br />
International Classification <strong>of</strong> Diseases (ICD-9) codes, 79<br />
intrahospital transportation <strong>of</strong> critically ill patients, 534,<br />
538<br />
ischemia, myocardial, 267, 271<br />
Joint Commission on the Accreditation <strong>of</strong> Healthcare<br />
Organizations (JCAHO)<br />
pain management guidelines, 396<br />
patient safety advocacy, 602<br />
physical restraint m<strong>and</strong>ates, 288<br />
root cause analysis m<strong>and</strong>ate, 51<br />
safety st<strong>and</strong>ards, 606<br />
sentinel event alerts, 47<br />
Sentinel Event Database, 43<br />
surveillance m<strong>and</strong>ate, 43<br />
wrong-site surgery, avoiding, recommendations, 495<br />
knee replacement surgery, VTE risk, 336<br />
laparoscopic cholecystectomy training<br />
complications with procedure, 214<br />
learning curve, 213<br />
research needs, 216<br />
skills needed, 213<br />
studies, 215<br />
latent error, 51<br />
Leapfrog Group<br />
high-volume centers, 205<br />
ICUs model recommendations, 413<br />
methods to promote patient safety practices, 607<br />
patient safety advocacy, 603<br />
referral system, 208<br />
learning curves for new procedures, 213<br />
legal issues<br />
end-<strong>of</strong>-life care, 556<br />
informed consent, 546<br />
legislation<br />
methods to promote patient safety practices, 607
Natural Death Acts, 556<br />
nursing issues, 425<br />
Omnibus Budget Reconciliation Act <strong>of</strong> 1987, 288, 302<br />
patient safety, 601<br />
<strong>Patient</strong> Self-Determination Act (PSDA), 556<br />
publication <strong>of</strong> performance data, 605<br />
Lerner’s Consumer Guide, 569<br />
lighting <strong>and</strong> fatigue, 524<br />
LINE/LOS checklist, 502<br />
living will for end-<strong>of</strong>-life care, 556<br />
localizing care to high-volume centers, 205<br />
low dose unfractionated heparin (LDUH) to prevent VTE,<br />
333, 335<br />
low molecular weight heparin (LMWH) to prevent VTE,<br />
333, 335<br />
machine-readable automatic identification (ID) systems<br />
conventional wrist b<strong>and</strong> use, 489<br />
described, 487<br />
medical record keeping, 489<br />
medication dispensing <strong>and</strong> administration, 488<br />
patient identification, 487<br />
patient identification studies, 490, 491<br />
research needs, 492<br />
specimen h<strong>and</strong>ling, 488<br />
malnutrition in hospitalized patients, 359, 363<br />
McLaughlin dispensing system, 111<br />
median sternotomy, 238<br />
medical device alarms, 462<br />
frequency reduction, 465<br />
research needs, 466<br />
source identification, 463<br />
urgency ranking, 463<br />
visual interfaces, 463<br />
medical devices<br />
design <strong>and</strong> development, 459<br />
device evaluation, 461, 462<br />
medical recordkeeping <strong>and</strong> bar codes, 489<br />
medical records, patient access to, 570<br />
medication dispensing <strong>and</strong> bar codes, 488<br />
medication dispensing systems, automated<br />
Baxter ATC-212 system, 111<br />
described, 111<br />
harmpotential, 113<br />
machine-readable code, 111<br />
McLaughlin system, 111<br />
research needs, 114<br />
studies, 111, 112, 115<br />
unit-dose distribution systems <strong>and</strong>, 112<br />
medication errors<br />
clinical pharmacists impact on, studies, 75<br />
dosage errors, 112<br />
identification <strong>of</strong> ADEs, 41<br />
information transfer between inpatient/outpatient<br />
pharmacies, 471<br />
studies, 106<br />
time-constraints <strong>and</strong>, 101<br />
unit-dose dispensing, 102<br />
MedTeams behavior-based teamwork system, 505<br />
melatonin supplementation for fatigue, 525<br />
methicillin-resistant Staphylococcus aureus (MRSA), 128<br />
minocycline/rifampin-coated catheters, 169<br />
minocycline-coated urinary catheters, 149<br />
659<br />
misidentification, avoiding<br />
bar coding, 487<br />
wrong-site surgery, 494, 496, 497<br />
wrong-site surgery, research needs, 498<br />
monitoring, intraoperative<br />
ASA st<strong>and</strong>ards, 268<br />
described, 265<br />
research needs, 267<br />
studies, 266<br />
napping, 524<br />
National Quality Forum (NQF), 21<br />
National Vaccine Advisory Committee, pneumococcal<br />
vaccines, 385<br />
Natural Death Acts, 556<br />
nausea due to PCA, 404<br />
neurosurgery, VTE risk, 337<br />
non-radiologists<br />
radiograph interpretations, 376<br />
training, 377, 381<br />
normal accident theory, 447<br />
normothermia, perioperative<br />
described, 231<br />
studies, 232, 233<br />
nosocomial infections<br />
antibiotics <strong>and</strong>, 141<br />
barrier precautions <strong>and</strong>, 127<br />
h<strong>and</strong>washing <strong>and</strong>, 119<br />
h<strong>and</strong>washing practices <strong>and</strong>, 120<br />
pathogens other than VRE <strong>and</strong> C. difficile, 128<br />
pneumonia <strong>and</strong> GI bleeding, 368<br />
nurse staffing<br />
availability studies, 426, 431<br />
care delivery models, 430<br />
care delivery studies, 426<br />
care practices evaluated, 424<br />
intervention studies, 427<br />
measures, 430<br />
nurse intervention <strong>and</strong> patient outcome studies, 437<br />
organization models <strong>and</strong> patient outcome studies, 435<br />
relationship <strong>of</strong> models to patient outcomes, 423<br />
research needs, 428<br />
studies, 425<br />
unions <strong>and</strong>, 425<br />
nutritional support practices<br />
described, 359<br />
studies, 360, 365<br />
Occupational <strong>Safety</strong> <strong>and</strong> Health Administration (OSHA),<br />
601<br />
open ICU model, 413<br />
opinion leaders for modifying physician behavior, 595, 597<br />
opioids, 404<br />
orientation techniques to prevent delirium, 307<br />
oropharyngeal secretions, 190<br />
osomolar iodinated contrast media to prevent RCIN, 349<br />
oxygen, perioperative<br />
described, 234<br />
future research needs, 236<br />
studies, 235, 236<br />
pain management<br />
abdomen, acute, 396<br />
about, 396<br />
acute pain services, described, 400
acute pain services, studies, 401, 403<br />
antiemtics for PCA, 404<br />
non-pharmacologic intervention studies, 407, 409<br />
non-pharmacologic interventions, 406<br />
patient education, 407<br />
research needs, 408<br />
palliative care programs, 561<br />
patient advocacy<br />
benefits, 569<br />
improving non-compliance with medical advice <strong>and</strong><br />
treatment, 570<br />
medical records, access to, 570<br />
research needs, 571<br />
patient classification systems (PCSs), 423<br />
patient mobility to prevent delirium, 307<br />
patient notification <strong>of</strong> abnormal results<br />
direct mail, 483<br />
problems <strong>and</strong> current methods, 482<br />
research needs, 484<br />
study, 483, 485<br />
<strong>Patient</strong> Self-Determination Act (PSDA), 556<br />
patient to nurse ratios, 424<br />
peer monitoring, 505<br />
percutaneous transluminal angioplasty, 271<br />
pericatheter skin inflammation, 163<br />
pharmacies<br />
automated drug dispensing, 102<br />
information transfer, 471, 473, 475<br />
pharmacologic measures to prevent venous<br />
thromboembolism, 333<br />
physical restraints for fall prevention in older patients<br />
described, 287<br />
studies, 288, 290<br />
physician behavior, modifying<br />
education techniques, research needs, 598<br />
education techniques, studies, 596, 599<br />
educational techniques, 595<br />
practice guidelines, 576<br />
pneumococcal infections<br />
re-infection risks, 385<br />
vaccines <strong>and</strong>, 385<br />
pneumococcal vaccines<br />
antibiotic resistance, 390<br />
delivery method studies, 392<br />
delivery methods, studies, 388<br />
described, 385<br />
effectiveness studies, 386, 387, 391<br />
research needs, 390<br />
pneumothorax, 247<br />
POLST (Physician Orders for Life-Sustaining Treatment),<br />
559, 564<br />
polymixin, topical, 193<br />
povidone-iodine (PI)<br />
central venous catheter insertion, 175<br />
compared to (CHG), studies, 177, 178<br />
practice guidelines for modifying physician behavior<br />
computer-based, 577<br />
described, 575<br />
research needs, 578<br />
studies, 576, 577<br />
pre-anesthesia equipment checklists<br />
described, 259<br />
660<br />
research needs, 261<br />
studies, 260, 262<br />
pressure ulcers<br />
in older patients, 301<br />
research needs, 304<br />
suprapubic catheters <strong>and</strong>, 156<br />
pr<strong>of</strong>essional societies <strong>and</strong> patient safety, 604<br />
pr<strong>of</strong>essional societies, to promote patient safety practices,<br />
607<br />
psychosocial support for pain management, 407<br />
public reporting systems, 605<br />
pulmonary embolization (PE), 333<br />
pulse oximetry, 265, 266, 466, 604<br />
Pyxis Medstations, 112<br />
radiocontrast-induced nephropathy (RCIN)<br />
described, 349<br />
prevention practices, 349<br />
prevention studies, 353, 354<br />
studies, 350, 351<br />
radiograph interpretation errors by non-radiologists<br />
described, 376<br />
digitized vs. film-based radiographs, 376<br />
reducing, 377, 378, 381<br />
research needs, 380<br />
radiologists <strong>and</strong> non-radiologists<br />
radiograph interpretations, 376<br />
training, 377<br />
reminder systems for modifying physician behavior, 595<br />
renal failure, RCIN <strong>and</strong>, 349<br />
report methodology<br />
abstraction elements, 34<br />
evidence review, 15<br />
exclusion criteria, 31<br />
exclusions, 16<br />
inclusion criteria, 31<br />
inclusions, 16<br />
organization, 18<br />
outcome measures, 32<br />
review methodologies, 29<br />
safety practice ratings, 613<br />
safety practices evaluation, 33<br />
safety practices organization, 29<br />
search strategy, 30<br />
study design hierarchy, 31<br />
review methodologies, error-based, 29<br />
rifampin-coated urinary catheters, 149<br />
risk assessment to prevent falls in older patients, 282<br />
root cause analysis (RCA)<br />
industrial accidents, 51<br />
JCAHO m<strong>and</strong>ate, 51<br />
process, 52<br />
research needs, 55<br />
studies, 53, 54<br />
safety culture, promoting<br />
accident theories, 447<br />
behavior-based programs, 450<br />
checklist <strong>of</strong> patient safety concepts, 453<br />
methods, medical <strong>and</strong> non-medical, 448<br />
research needs, 451<br />
VHA initiative, 449<br />
safety practice ratings<br />
data inputs, 613
decision rules <strong>and</strong> judgment considerations, 615<br />
goals, 613<br />
impact <strong>and</strong> evidence overall rating, 616<br />
rating process, 614<br />
rating tables, 617<br />
research priority overall rating, 616<br />
safety practices<br />
all practices, categorical ratings, <strong>and</strong> comments, 633<br />
clinical vs. general issues in content, 17<br />
definition <strong>and</strong> scope, 13, 29<br />
further research likely to be beneficial, 630<br />
further research likely to be highly beneficial, 627<br />
greatest strength <strong>of</strong> evidence regarding their impact <strong>and</strong><br />
effectiveness, 620<br />
high strength <strong>of</strong> evidence regarding their impact <strong>and</strong><br />
effectiveness, 621<br />
hospital care <strong>and</strong>, 14<br />
industries outside health care, 25<br />
low strength <strong>of</strong> evidence regarding their impact <strong>and</strong><br />
effectiveness, 624<br />
lowest strength <strong>of</strong> evidence regarding their impact <strong>and</strong><br />
effectiveness, 625<br />
medium strength <strong>of</strong> evidence regarding their impact <strong>and</strong><br />
effectiveness, 622<br />
organization <strong>of</strong>, 29<br />
rated by strength <strong>of</strong> evidence, 619<br />
research priority ratings, 627<br />
selective digestive tract decontamination (SDD)<br />
described, 193<br />
studies, 194, 196<br />
semi-recumbent positioning<br />
described, 185<br />
studies, 186, 189<br />
Sentinel Event Database, 43<br />
shift rotation <strong>and</strong> fatigue, 523<br />
sign-out systems for cross-coverage<br />
described, 476<br />
research needs, 478<br />
studies, 477, 479<br />
silicone urethral catheters, 149<br />
silver alloy catheters<br />
bacteriuria reduction, 151<br />
described, 149<br />
simulator-based training<br />
advantages, 511<br />
anesthesia, 512<br />
cardiology, 514<br />
CRM <strong>and</strong>, 512<br />
gastroenterology, 514<br />
radiology, 513<br />
research needs, 514<br />
risks, 515<br />
surgery, 513<br />
sleep deprivation<br />
lighting at work, 524<br />
medical errors <strong>and</strong>, 519<br />
medical therapies speed, 525<br />
napping, 524<br />
research needs, 526<br />
shift limits <strong>and</strong> error reduction, 521<br />
shift rotation <strong>and</strong>, 520<br />
shift rotation direction <strong>and</strong> speed, 523<br />
661<br />
sleep hygiene education, 523<br />
studies, 522<br />
sponge, sharp, <strong>and</strong> instrument counts<br />
described, 255<br />
research needs, 256<br />
studies, 256<br />
staff cohorting as a barrier precaution, 127<br />
staphylococci <strong>and</strong> catheter-related infections, 163<br />
sternal wound infections, 238<br />
streptococcus pneumoniae, 385<br />
stress ulcers<br />
H2-receptor antagonist therapy, 198<br />
studies, 201, 202<br />
subclavian-vein catheters, 246<br />
subglottic secretions, continuous aspiration <strong>of</strong><br />
described, 190<br />
diagram, 192<br />
studies, 193<br />
sucralfate<br />
gastrointestinal bleeding reduction, described, 368<br />
gastrointestinal bleeding reduction, studies, 369, 372<br />
studies, 201, 202<br />
VAP reduction, 198<br />
suprapubic catheters<br />
compared to st<strong>and</strong>ard catheters, 156<br />
compared to urethral, studies, 159<br />
placement <strong>of</strong>, 155<br />
research needs, 157<br />
studies, 157<br />
surgical site infections (SSI)<br />
antimicrobial prophylaxis <strong>and</strong>, 221<br />
blood flow <strong>and</strong> body temperature, 231<br />
glucose control for diabetics <strong>and</strong>, 237<br />
normothermia <strong>and</strong>, 232<br />
oxygen <strong>and</strong>, 234<br />
research needs, 223, 233<br />
studies, 222, 224, 226<br />
surgical sponges <strong>and</strong> instruments, retained, 255<br />
teicoplanin <strong>and</strong> central venous catheters, 179<br />
teleradiology, 376<br />
thromboembolism<br />
heparin <strong>and</strong> warfarin monitoring, 88<br />
preventions, 333<br />
To Err is Human: Building a Safer Health System, 13, 25<br />
tobramycin, topical, 193<br />
total parental nutrition (TPN), 359, 361<br />
training<br />
end-<strong>of</strong>-life care, for physicians, 560<br />
laparoscoopic cholecystectomy, 213<br />
nurses, 424<br />
radiograph interpretations, 377<br />
simulator-based, 511<br />
transesophageal echocardiography, 265<br />
transfusion errors, 489<br />
transport teams, 534<br />
transportation <strong>of</strong> critically ill patients<br />
about, 534<br />
interhospital, 534<br />
intrahospital transport, 534<br />
manual vs. mechanical ventilation, studies, 541<br />
transport teams vs. st<strong>and</strong>ard, studies, 540<br />
transporting children, 535, 537
transurethral resection <strong>of</strong> the prostate, 157<br />
trauma patients, VTE risk, 338<br />
UCSF-Stanford Evidence-based Practice Center (EPC), 13<br />
ultrasound guidance <strong>of</strong> central vein catheterization<br />
described, 245<br />
insertion success rates, 246<br />
research needs, 249<br />
studies, 246, 247, 250<br />
unions <strong>and</strong> nurse staffing, 425<br />
unit-dose dispensing<br />
automated dispensing systems, 112<br />
described, 101<br />
dispensing variations, 101<br />
JCAHO st<strong>and</strong>ards, 102<br />
research needs, 105<br />
safety targets, 102<br />
studies, 103, 104, 106<br />
urethral catheters<br />
compared to suprapubic, 159<br />
latex, 149<br />
patient discomfort, 156<br />
research needs, 157<br />
silicone, 149<br />
urethral strictures, 157<br />
urinary catheters<br />
impregnated with antibiotic, 149<br />
silver alloy catheters, 149<br />
studies, 150, 153<br />
urinary tract infections (UTIs)<br />
catheters <strong>and</strong>, 149<br />
prevalence, 149<br />
research needs, 151<br />
usability testing, 461<br />
vancomycin, systemic use with central venous catheters,<br />
179<br />
vancomycin-resistant enterococci (VRE)<br />
antibiotic studies, 142, 143<br />
antibiotics <strong>and</strong>, 141<br />
barrier precaution studies, 129<br />
barrier precautions <strong>and</strong>, 127<br />
C. difficile relationship, 128<br />
research needs, 131, 144<br />
transferability, 127<br />
venography, 334<br />
662<br />
venous thromboembolism (VTE)<br />
described, 333<br />
preventative measures, mechanical <strong>and</strong> pharmacologic,<br />
343<br />
preventions, 333<br />
prophylaxis for surgical procedures <strong>and</strong> medical<br />
conditions, 345<br />
risk in general surgery, 335<br />
risk in hospitalized medical patients, 339<br />
risk in neurosurgery, 337<br />
risk in orthopedic surgery, 336<br />
risk reduction methods, 344<br />
studies, 334<br />
therapy factors, 337<br />
urethral catheters <strong>and</strong>, 156<br />
ventilation when transporting critically ill patients, 538,<br />
541<br />
ventilator-associated pneumonia (VAP)<br />
continuous aspiration <strong>of</strong> subglottic secretions, 190<br />
continuous oscillation, 185, 187<br />
described, 185<br />
patient positioning studies, 189<br />
research needs, 195, 199, 203<br />
SDD, 193<br />
semi-recumbent positioning, 185, 186<br />
sucralfate, 198<br />
Veterans Health Administration (VHA), 449<br />
vomiting due to PCA, 404<br />
ward stock system, 102<br />
warfarin<br />
about, 87<br />
dosing <strong>and</strong> monitoring errors, reducing, 87<br />
monitoring, 88<br />
self-management, studies, 89, 96<br />
venous thromboembolism prevention, 333<br />
warfarin nomograms, studies, 92<br />
warming blankets for normothermia during surgery, 231<br />
web-based health information, 16, 292, 451, 561, 570<br />
web-based programs, 478<br />
wrong-site surgery, avoiding<br />
described, 494<br />
research needs, 498<br />
studies, 496, 497